Solar blind ultraviolet imaging light spot analysis method and system

Abstract

The invention relates to the technical field of ultraviolet imaging detection, in particular to a solar blind ultraviolet imaging light spot analysis method and system. The method comprises the following steps: based on different gains of a solar blind imager, pre-establishing ANFI S evaluation models corresponding to the different gains for evaluating the discharge intensity based on a light spot image; s1, acquiring a light spot image of the surface of the electrical equipment; s2, performing feature extraction on the light spot image, and determining an emission radiation area s and an observation distance d of the light spot image; and S3, based on the gain corresponding to the solar blind imager, inputting the emission radiation area s and the observation distance d into the ANFI S evaluation model of the corresponding gain, so as to evaluate the predicted discharge amount q on the surface of the electrical equipment. The system comprises a solar-blind imager used for collecting light spot images so as to carry out solar-blind ultraviolet imaging light spot analysis. According to the method, factors such as the discharge radiation area, the discharge capacity, the observation distance and the imager gain are comprehensively considered, the ANFI S algorithm is applied, an accurate estimation model is constructed, and accurate estimation of the corona discharge capacity is achieved.

Description

Technical Field

[0001] The present application relates to the field of ultraviolet imaging detection technology, and more specifically, to a solar-blind ultraviolet imaging spot analysis method and system. Background Art

[0002] Under long-term operation conditions, high-voltage electrical equipment may experience problems such as reduced insulation performance, structural damage, and surface contamination, which may cause electric field concentration and thus corona discharge, affecting the safe operation of the equipment.

[0003] The invention patent with announcement number CN102305903B discloses a pulse current partial discharge detection method on a live single-phase integrated GIS. The pulse current test method is used to perform partial discharge detection on a live single-phase integrated GIS without power outage and disconnection. Based on the pulse current discharge detection method, the partial discharge signal inside the device is reflected in the voltage change V of the detection impedance, and the discharge amount of the device is determined by measuring the change in voltage. However, due to the presence of more electromagnetic pulse interference at the actual substation site, these pulse signals will affect the final measurement results.

[0004] The spectrum analysis of the equipment discharge phenomenon shows that its wavelength range is 230nm~405nm, and the wavelength range of ultraviolet light is 40nm~400nm. Sunlight also contains ultraviolet light. Because the earth's ozone layer absorbs part of the wavelength component, the spectrum segment of 240nm~280nm is called the solar blind zone, and the amount of ultraviolet light emitted by the sun in this segment is extremely low. According to this characteristic, an ultraviolet imager for daytime corona detection has been developed. By receiving the signal generated by corona discharge, the image is processed and superimposed on the visible light image to determine the position of the corona and the intensity of the discharge, providing a basis for the status detection of the equipment. Corona discharge will produce ultraviolet light. Even weaker discharge can be observed by ultraviolet imaging instruments. Ultraviolet imaging technology has been widely used in the detection test of corona discharge of high-voltage electrical equipment in power systems, live operation and maintenance, using light in the ultraviolet spectrum range to obtain images, and determining the location of the discharge and judging the intensity of the discharge based on the ultraviolet imaging spectrum.

[0005] The invention patent with announcement number CN117607636B discloses a multi-spectral fusion integrated sensing, transmission, storage and computing high-voltage discharge detection method. S1: The visible light image and ultraviolet image of the equipment are respectively annotated by the YOLOv3 algorithm to obtain the equipment and ultraviolet discharge spot image annotation boxes; S2: Determine whether there is an ultraviolet discharge spot in the spot image annotation box. If so, generate a high-voltage partial discharge abnormal alarm message, otherwise enter S4; S3: Determine whether there are other discharge devices in the device annotation box generated based on the visible light image of the equipment. If so, adjust the camera angle of the detection device, shoot the high-voltage discharge equipment and generate an alarm message, otherwise enter S4; S4: Adjust the local data acquisition frame rate based on the size of the ultraviolet discharge spot annotation box and the queue backlog; S5: Establish multiple sensing, transmission, storage and computing schemes, monitor the changes in the communication environment in real time, and perform adaptive shearing of the sensing, transmission, storage and computing schemes; S6: Dynamically allocate bandwidth to ensure the reliable transmission of alarm messages and image data. Based on the technical solution provided by the patent, the network diagnosed the larger discharge spots of the insulators in operation on site and gave accurate evaluation results, but ignored the relatively smaller discharge spots around them, which needs to be improved.

[0006] Therefore, how to provide a discharge detection method that is not disturbed by the electromagnetic pulses existing at the substation site and does not ignore the smaller light spots around it, and can achieve visual and accurate identification of the operating status of high-voltage electrical equipment, has become a technical problem that needs to be solved. Summary of the invention

[0007] In view of this, the present application provides a day-blind ultraviolet imaging spot analysis method and system, which comprehensively considers factors such as discharge radiation area, discharge amount, observation distance and imager gain, and can be applied to corona discharge analysis of different scenarios and equipment to achieve accurate analysis of the light spot of electrical equipment.

[0008] The technical solutions provided by this application are as follows:

[0009] In a first aspect, the present application provides a solar-blind ultraviolet imaging spot analysis method, comprising:

[0010] S1. Obtaining a spot image on the surface of the electrical equipment collected by a solar-blind imager;

[0011] S2, extracting features from the spot image to determine the emission radiation area s and observation distance d of the spot image;

[0012] S3. Based on the gain corresponding to the solar-blind imager, the extracted emission radiation area s and observation distance d are input into the ANFIS evaluation model of the corresponding gain to evaluate the predicted discharge amount q on the surface of the electrical equipment.

[0013] In one possible implementation, the method before S1 further includes: based on multiple gains of the solar-blind imager, pre-establishing an adaptive neural fuzzy inference system ANFIS evaluation model corresponding to each gain, including:

[0014] Obtain training sample data and establish a data set of observation distance d, emission radiation area s and discharge intensity q corresponding to each gain;

[0015] Determine the number of fuzzy subsets of each input variable of the ANFIS evaluation model to be established, and select a Gaussian function as a membership function; wherein the input variables include an observation distance d and an emission radiation area s;

[0016] Based on the input variables, the grid segmentation method is used to automatically establish the initial fuzzy reasoning, the number of rules is set to 100, and a linear combination of the input variables and a constant term is obtained. The calculation formula of each rule is expressed as:

[0017] y i =a0 i +a1 i x1+…+a r i x r

[0018] In the formula, y i is the output of the ith rule; x is the input variable; а is the coefficient;

[0019] Based on the calculation formula of each rule, the final output expression is obtained, which is expressed as:

[0020]

[0021] In the formula, O i is the final output, w i is the trigger weight of each rule, calculated by the T-norm operator.

[0022] In one possible implementation, after obtaining the training sample data, the method further includes:

[0023] Based on the training sample data, determining the relationship between the emission radiation area s and the observation distance d under the same discharge intensity;

[0024] The data sets of observation distance, emission radiation area and discharge intensity under different discharge intensities are established respectively, which are expressed as (d, s, q).

[0025] In one possible implementation, after obtaining the final output expression, the method further includes:

[0026] A hybrid training algorithm is used to repeatedly train the ANFIS evaluation model to adjust the parameters of the membership function.

[0027] In one possible implementation, after establishing the data set of the observation distance d, the emission radiation area s and the discharge intensity q, the method further includes:

[0028] According to a preset ratio, the data set is divided into training data and test data; the training data is used to train the ANFIS evaluation model, and the test data is used to input into the trained ANFIS evaluation model to obtain the root mean square error of the test data and verify the prediction accuracy of the ANFIS evaluation model.

[0029] In one possible implementation, the gain of the solar-blind imager is set to a fixed gain of 50%, 60%, 70%, and 80%, and ANFIS evaluation models corresponding to the fixed gains of 50%, 60%, 70%, and 80% are established respectively to evaluate the discharge intensity on the surface of electrical equipment.

[0030] In one possible implementation manner, after acquiring the spot image in S1, the method further includes:

[0031] Based on image processing technology, the light spot image is enhanced, denoised and segmented to extract the light spot information.

[0032] In one possible implementation, the feature extraction of the spot image in S2 includes:

[0033] The light spot image is subjected to edge detection, segmentation and shape description to obtain the size, shape and position characteristics of the light spot, which are used to determine the emission radiation area s and observation distance d of the light spot.

[0034] In one possible implementation manner, after S3, the method further includes:

[0035] Obtain historical fault samples and related characteristic parameters of electrical equipment and establish a fault classification model; the related characteristic parameters include light spot characteristic parameters on the surface of the electrical equipment and the corresponding discharge amount; the light spot characteristic parameters include the emission radiation area s and the observation distance d;

[0036] According to the current spot image on the surface of the electrical equipment and the predicted discharge amount obtained by evaluation, based on the fault evaluation model, the fault type of the electrical equipment is determined, including thermal fault, poor contact and insulation damage.

[0037] In a second aspect, the present application provides a solar-blind ultraviolet imaging spot analysis system, which is used to perform the solar-blind ultraviolet imaging spot analysis method as described in the first aspect, comprising:

[0038] A solar-blind imager, including a sensor, a lens and an imaging module, is used to collect ultraviolet signals of 240-280 nanometers and obtain a spot image on the surface of electrical equipment;

[0039] A model building module is used to pre-build an adaptive neural fuzzy inference system ANFIS evaluation model corresponding to different gains based on different gains of the solar-blind imager to evaluate the discharge intensity;

[0040] A feature extraction module is used to extract features of the spot image and determine the emission radiation area s and observation distance d of the spot image;

[0041] The evaluation module is used to input the extracted emission radiation area s and observation distance d into the ANFIS evaluation model of the corresponding gain based on the gain corresponding to the solar-blind imager, so as to evaluate the predicted discharge amount q on the surface of the electrical equipment.

[0042] Compared with the prior art, the technical solution of this application has the following beneficial effects:

[0043] This application proposes an innovative quantification method, which explores the complex relationship between discharge radiation area, discharge amount, observation distance and imager gain, creatively uses the adaptive neural fuzzy inference system ANFIS algorithm, and successfully builds an accurate estimation model. This model can effectively analyze the physical characteristics of the corona discharge process and achieve accurate estimation of the corona discharge amount.

[0044] This application can be applied to corona discharge analysis of different scenarios and equipment by comprehensively considering factors such as discharge radiation area, discharge amount, observation distance and imager gain, greatly broadening its application field. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0046] Figure 1 A flow chart of a solar-blind ultraviolet imaging spot analysis method provided in Example 1 of the present application.

[0047] Figure 2 A flowchart of a method for training an adaptive neural-fuzzy inference system ANFIS evaluation model provided in Example 1 of the present application.

[0048] Figure 3This is a relationship curve between the emitted radiation area s and the observation distance d when the gain is 70% provided in Example 2 of the present application.

[0049] Figure 4 A schematic diagram of comparing actual values ​​and predicted values ​​for model performance verification provided in Example 2 of the present application.

[0050] Figure 5 This is a schematic diagram of the structure of a solar-blind ultraviolet imaging spot analysis system provided in Example 3 of the present application. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0052] Based on the above problems, the present application provides a day-blind ultraviolet imaging spot analysis and diagnosis method based on an adaptive neurofuzzy inference system (ANFIS) algorithm. ANFIS is an adaptive inference system that integrates fuzzy logic and neural networks. By applying the ANFIS algorithm to day-blind ultraviolet imaging technology, the accuracy and reliability of the analysis of the light spot of electrical equipment can be improved. Traditional spot analysis methods often rely on empirical rules or manually selected features, while ANFIS can automatically learn and adjust parameters, thereby improving the accuracy and reliability of spot analysis.

[0053] Solar-blind ultraviolet imaging technology is a non-destructive testing method that can obtain the surface temperature distribution of electrical equipment under sunlight. This technology uses ultraviolet radiation emitted from the surface of electrical equipment to capture and analyze the spot image through imaging equipment, and can monitor the operating status and fault conditions of the equipment in real time. Compared with traditional infrared imaging technology, solar-blind ultraviolet imaging technology can better suppress sunlight interference and improve imaging quality.

[0054] The technical solution provided by this application combines ultraviolet imaging technology with the ANFIS algorithm. A solar-blind imager is used to collect the spot image on the surface of the electrical equipment. Corresponding to the different gains of the solar-blind imager, ANFIS evaluation models are constructed respectively to achieve accurate analysis and fault diagnosis of the spot of the electrical equipment, improve the accuracy and reliability of the spot analysis, and at the same time be able to deal with problems such as sunlight interference that plague traditional infrared imaging technology.

[0055] Below, a day-blind ultraviolet imaging spot analysis method provided by the present application will be described in detail in conjunction with specific embodiments.

[0056] Embodiment 1:

[0057] See also Figure 1 , is a flow chart of a solar-blind ultraviolet imaging spot analysis method provided in Example 1 of the present application. Figure 1 As shown in , the specific implementation steps of the above method may include:

[0058] S1. Acquire a light spot image on the surface of electrical equipment.

[0059] Corona discharge radiates light signals. In the weak discharge stage, the light radiation energy is mainly distributed in the ultraviolet band of 200-400nm, which is invisible to the human eye. The solar-blind imager only detects ultraviolet signals of 240-280 nanometers, so it can avoid interference from sunlight. Compared with traditional discharge detection methods, the ultraviolet imaging method has many advantages, such as non-contact, strong anti-interference ability, and high positioning accuracy.

[0060] Establishing the relationship between the UV signal and the electrical signal can help us estimate the discharge intensity using the UV imaging method. To quantify the discharge intensity, some quantitative parameters are needed. Currently, the "photon number" parameter is widely used, but it has a complex relationship with the gain of the imager, so it is still difficult to quantify the discharge intensity. Existing experiments show that the discharge UV radiation changes with the discharge amount. When the discharge intensity increases, the air molecules will be ionized in a larger space, radiating more UV signals, and the discharge image will also increase. The emitted radiation has a clear physical meaning, so image parameters are used to quantify the discharge intensity.

[0061] In the embodiment of the present application, the spot image on the surface of the electrical equipment is captured by optical imaging technology, including sensors, lenses, and imaging systems. The electrical equipment includes generators, transformers, and cables. By real-time detection and analysis of the spot image, thermal failures, poor contact, and insulation damage of the electrical equipment can be discovered early, thereby improving the reliability and safety of the equipment.

[0062] S2. Extract features from the light spot image to determine the emission radiation area s and observation distance d of the light spot image.

[0063] Based on the above research status, this application provides a new quantification method, that is, through the relationship between the emission radiation area and the discharge amount, the observation distance and the imager gain, an estimation model is established using the ANFIS algorithm to achieve the estimation of the corona discharge amount.

[0064] The emission radiation area is related to the discharge amount, observation distance and imager gain. The problem of estimating the discharge amount is actually the process of predicting the discharge amount q based on the emission radiation area s, observation distance d and gain g. It can be regarded as a multidimensional nonlinear mapping function f, which can be expressed as

[0065] q=f(s,d,g) (1)

[0066] The relationship between q and s, d, and g is very complex, and it is difficult to give a specific function expression. The ANFIS algorithm has good nonlinear mapping capabilities. Using the Matlab fuzzy logic toolbox, the discharge level estimation model was established using the ANFIS algorithm.

[0067] Directly using the emission radiation area s, observation distance d, and gain g to predict the flow size, the ANFIS model has three input variables, and the model structure is complex. Practical application shows that when the gain is set to 50%, 60%, 70%, and 80% of the evaluation model, for each model, the input variables are simplified to two parameters: observation distance d and emission radiation area s, and the output is the predicted discharge amount q.

[0068] S3. Based on the gain corresponding to the solar-blind imager, the extracted emission radiation area s and observation distance d are input into the ANFIS evaluation model of the corresponding gain to evaluate the predicted discharge amount q on the surface of the electrical equipment.

[0069] Compared with the prior art, the technical solution of this application has the following beneficial effects:

[0070] This application proposes an innovative quantification method, which explores the complex relationship between discharge radiation area, discharge amount, observation distance and imager gain, creatively uses the adaptive neural fuzzy inference system ANFIS algorithm, and successfully builds an accurate estimation model. This model can effectively analyze the physical characteristics of the corona discharge process and achieve accurate estimation of the corona discharge amount.

[0071] By comprehensively considering factors such as discharge radiation area, discharge amount, observation distance and imager gain, corona discharge analysis applicable to different scenarios and equipment can be realized, greatly broadening its application areas.

[0072] In practical scenarios, this technology can be applied to the spot analysis and fault diagnosis of various electrical equipment, such as generators, transformers, cables, etc. By real-time monitoring and analysis of the spot image, problems such as thermal failure, poor contact and insulation damage of the equipment can be discovered early, thus improving the reliability and safety of the equipment. In addition, this technology can also be applied to the inspection and maintenance of the power system to improve the operating efficiency and sustainable development capabilities of the power system.

[0073] Based on pattern recognition technology, classification and diagnosis are performed according to the characteristic parameters of the light spot, including machine learning, fuzzy reasoning, statistical analysis and other methods. Fault diagnosis technology is used to judge the operating status and existing faults of the equipment based on the light spot analysis results, including fault model establishment, fault feature extraction and fault classification and other methods.

[0074] Obtain historical fault samples and related characteristic parameters of electrical equipment and establish a fault classification model. The above-mentioned related characteristic parameters include the light spot characteristic parameters on the surface of the electrical equipment and the corresponding discharge amount. The light spot characteristic parameters include the emission radiation area s and the observation distance d.

[0075] According to the current spot image on the surface of the electrical equipment and the predicted discharge amount obtained by evaluation, based on the fault evaluation model, the fault type of the electrical equipment is determined, including the above-mentioned thermal fault, poor contact and insulation damage.

[0076] Embodiment 2:

[0077] Since the discharge magnitudes are close under the same emission radiation area, but it is difficult to determine the discharge type in actual detection, we can use the burr model data to estimate the discharge magnitude approximately. The second embodiment of the present application takes the construction of an ANFIS evaluation model with a corresponding gain of 70% as an example.

[0078] See also Figure 2 , which is a flowchart of a method for training an adaptive neural-fuzzy inference system ANFIS evaluation model provided in Example 2 of the present application.

[0079] (1) Determination of training sample data. Figure 3 , which is a relationship curve between the emitted radiation area s and the observation distance d when the gain is 70% provided in Example 2 of the present application.

[0080] like Figure 3 As shown in , each curve represents the relationship between the emission radiation area and the distance under the same discharge intensity. Therefore, based on the experimental data, we can establish a data set of observation distance, emission radiation area and discharge intensity. The data set is expressed in the form of (d, s, q). For example, for the curve of the discharge magnitude 4013, at seven typical distances, the data sets are (4,22137,4013), (8,8157,4013), (16,2022,4013), (24.5,917,4013)), (29.6,632,4013), (33,426,4013), (41.5,282,4013). Similarly, data sets can also be established for other discharge intensities. 784 sets of sample data were obtained in the experiment, half of which were selected as training data and the rest were selected as test data.

[0081] (2) Fuzzy partitioning of input variables and selection of membership functions

[0082] To establish the ANFIS model, it is necessary to determine the number of fuzzy subsets for each input variable and then select the appropriate membership function. Considering the prediction accuracy and training time cost, the number of fuzzy subsets for distance and emission radiation area is set to 10, and Gaussian function is selected as the membership function.

[0083] (3) Establishment of fuzzy rules

[0084] Call the "genfis1" function in the fuzzy logic toolbox to automatically establish the initial fuzzy reasoning based on the input data using the grid segmentation method. The total number of rules is 100. The rules are represented by a series of if-then statements. Each output is a linear combination of the input variables and a constant term. The calculation formula for each rule is as follows:

[0085] y i =a0 i +a1 i x1+…+a r i x r (2)

[0086] Where: yi is the output of the ith rule; x is the input variable; a is the coefficient. Since the total number of rules is 100, there are 100 output expressions in total. The final output expression is formula (3).

[0087]

[0088] Among them: i is the final output; w i is the trigger weight of each rule, which is calculated by the T-norm operator, generally using minimization operation or algebraic product operation.

[0089] (4) Training of prediction models

[0090] In the initial system generated by the genfis1 function, the parameters of the membership function are not optimal; they need to be adjusted through training. Here, a hybrid training algorithm is used, the number of training times is set to 50, and the entire training time is 68s.

[0091] (5) Model performance verification

[0092] For a prediction model, accuracy and generalization ability are very important. When the training data and test data are input into the trained model, the true value and the predicted value are as follows. Figure 4 shown.

[0093] Embodiment 3

[0094] See also Figure 5, is a schematic diagram of the structure of a solar-blind ultraviolet imaging spot analysis system provided in an embodiment of the present application. Figure 5 As shown in , the solar-blind ultraviolet imaging spot analysis system includes a solar-blind imager 1, a model building module 2, a feature extraction module 3 and an evaluation module 4.

[0095] The solar blind imager 1 is based on optical imaging technology and is used to capture the light spot image on the surface of electrical equipment. Specifically, it includes a sensor, a lens, and an imaging system.

[0096] The model building module 2 is specifically used to pre-establish an adaptive neural fuzzy inference system ANFIS evaluation model corresponding to different gains based on different gains of the solar-blind imager to evaluate the discharge intensity.

[0097] The feature extraction module 3 is specifically used to extract features from the spot image and determine the emission radiation area s and observation distance d of the spot image.

[0098] Image processing technology is used to enhance, denoise, segment and extract features of the light spot image to extract useful information. Light spot feature extraction involves edge detection, segmentation and shape description of the light spot image to obtain information such as the size, shape and position characteristics of the light spot.

[0099] The evaluation module 4 is specifically used to input the extracted emission radiation area s and observation distance d into the ANFIS evaluation model of the corresponding gain based on the gain corresponding to the solar-blind imager, so as to evaluate the predicted discharge amount q on the surface of the electrical equipment.

[0100] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0101] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0102] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A solar-blind ultraviolet imaging spot analysis method, characterized in that: The method comprises: S1. Obtaining a spot image on the surface of the electrical equipment collected by a solar-blind imager; S2, extracting features from the spot image to determine the emission radiation area s and observation distance d of the spot image; S3. Based on the gain corresponding to the solar-blind imager, the extracted emission radiation area s and observation distance d are input into the adaptive neural fuzzy inference system ANFIS evaluation model corresponding to the gain, so as to evaluate the predicted discharge amount q on the surface of the electrical equipment.

2. A solar-blind ultraviolet imaging spot analysis method according to claim 1, characterized in that: Before step S1, the method further includes: pre-establishing an adaptive neural fuzzy inference system ANFIS evaluation model corresponding to each gain based on multiple gains of the solar-blind imager, including: Obtain training sample data and establish a data set of observation distance d, emission radiation area s and discharge intensity q corresponding to each gain; Determine the number of fuzzy subsets of each input variable of the ANFIS evaluation model to be established, and select a Gaussian function as a membership function; wherein the input variables include an observation distance d and an emission radiation area s; Based on the input variables, the grid segmentation method is used to automatically establish the initial fuzzy reasoning, the number of rules is set to 100, and a linear combination of the input variables and a constant term is obtained. The calculation formula of each rule is expressed as: and i =a0 i +a1 i x1+…+a r i x r In the formula, y i is the output of the ith rule; x is the input variable; а is the coefficient; Based on the calculation formula of each rule, the final output expression is obtained, which is expressed as: In the formula, O i is the final output, w i is the trigger weight of each rule, calculated by the T-norm operator.

3. A solar-blind ultraviolet imaging spot analysis method according to claim 2, characterized in that: After obtaining the training sample data, the method further includes: Based on the training sample data, determining the relationship between the emission radiation area s and the observation distance d under the same discharge intensity; The data sets of observation distance, emission radiation area and discharge intensity under different discharge intensities are established respectively, which are expressed as (d, s, q).

4. A solar-blind ultraviolet imaging spot analysis method according to claim 2, characterized in that: After obtaining the final output expression, the method further includes: A hybrid training algorithm is used to repeatedly train the ANFIS evaluation model to adjust the parameters of the membership function.

5. A solar-blind ultraviolet imaging spot analysis method according to claim 2, characterized in that: After establishing the data set of observation distance d, emission radiation area s and discharge intensity q, the method further comprises: According to a preset ratio, the data set is divided into training data and test data; the training data is used to train the ANFIS evaluation model, and the test data is used to input into the trained ANFIS evaluation model to obtain the root mean square error of the test data and verify the prediction accuracy of the ANFIS evaluation model.

6. A solar-blind ultraviolet imaging spot analysis method according to claim 1, characterized in that: The gain of the solar-blind imager was set to fixed gains of 50%, 60%, 70%, and 80%. ANFIS evaluation models corresponding to fixed gains of 50%, 60%, 70%, and 80% were established to evaluate the discharge intensity on the surface of electrical equipment.

7. A solar-blind ultraviolet imaging spot analysis method according to claim 1, characterized in that: After acquiring the spot image in S1, the method further includes: Based on image processing technology, the light spot image is enhanced, denoised and segmented to extract the light spot information.

8. A solar-blind ultraviolet imaging spot analysis method according to claim 1, characterized in that: The feature extraction of the spot image in S2 includes: The light spot image is subjected to edge detection, segmentation and shape description to obtain the size, shape and position characteristics of the light spot, which are used to determine the emission radiation area s and observation distance d of the light spot.

9. A solar-blind ultraviolet imaging spot analysis method according to claim 1, characterized in that: After S3, the method further includes: Obtain historical fault samples and related characteristic parameters of electrical equipment and establish a fault classification model; the related characteristic parameters include light spot characteristic parameters on the surface of the electrical equipment and the corresponding discharge amount; the light spot characteristic parameters include the emission radiation area s and the observation distance d; According to the current spot image on the surface of the electrical equipment and the predicted discharge amount obtained by evaluation, based on the fault evaluation model, the fault type of the electrical equipment is determined, including thermal fault, poor contact and insulation damage.

10. A solar-blind ultraviolet imaging spot analysis system, used to execute the solar-blind ultraviolet imaging spot analysis method according to any one of claims 1 to 9, characterized in that: include: A solar-blind imager, including a sensor, a lens and an imaging module, is used to collect ultraviolet signals of 240-280 nanometers and obtain a spot image on the surface of electrical equipment; A model building module is used to pre-build an adaptive neural fuzzy inference system ANFIS evaluation model corresponding to different gains based on different gains of the solar-blind imager to evaluate the discharge intensity; A feature extraction module is used to extract features of the spot image and determine the emission radiation area s and observation distance d of the spot image; The evaluation module is used to input the extracted emission radiation area s and observation distance d into the ANFIS evaluation model of the corresponding gain based on the gain corresponding to the solar-blind imager, so as to evaluate the predicted discharge amount q on the surface of the electrical equipment.

Images