Optical push-broom positioning method for reactor insulation faults

Abstract

The optical push-scan positioning method for the reactor insulation fault comprises the following steps: supporting a beam to periodically push-scan under the actuation of a transmission device; collecting the light intensity distribution of a circular area on the bottom plane of the reactor under different light sources, extracting the gray image of the circular area, and obtaining the data matrix of the circular area; based on image transformation, the data matrix of the circular area is unfolded along the radial direction, and is converted into a rectangular data matrix through data filling; after unfolding, the circumference of the circle becomes the length of the rectangle, the radius of the circle becomes the width of the rectangle, the data of the circle center is changed into the length of the other side of the rectangle through data replication, and the data matrix of the circular area becomes the rectangular data matrix, and each column of the rectangular data matrix is the data along the radius of the data matrix of the circular area; based on the principal component analysis method, the data of the rectangular data matrix is analyzed to extract feature points, and multiple SiC ultraviolet solar blind sensor probes are linearly and non-uniformly arranged on the supporting beam according to the positions of the feature points.

Description

Technical Field

[0001] This invention relates to the field of online monitoring of reactors, and in particular to an optical push-broom method for locating insulation faults in reactors. Background Technology

[0002] Reactors are crucial equipment in power systems, serving functions such as compensating for stray capacitive currents, limiting inrush currents, limiting short-circuit currents, filtering, and wave blocking. In current power systems, reactors are used extensively. However, due to their structural characteristics, they are susceptible to partial discharge, insulation damage, inter-turn short circuits, and other faults, as well as threats posed by insecure fastener installation, all of which can affect the stable operation of the power grid.

[0003] The information disclosed in the background section is only intended to enhance the understanding of the background of the present invention, and therefore may contain information that does not constitute prior art known to those skilled in the art in this country. Summary of the Invention

[0004] To address the shortcomings or defects of the existing technology, an optical push-broom method for locating insulation faults in reactors is provided. This method can conveniently detect potential fault threats in dry-type reactors during inspections, while avoiding electromagnetic interference on-site and meeting sensitivity requirements. It can detect whether the reactor is discharging and locate the position of key discharge points.

[0005] The objective of this invention is achieved through the following technical solutions.

[0006] An optical push-broom method for locating insulation faults in reactors includes the following steps:

[0007] The base is located inside the reactor, and the center of the base is collinear with the central axis of the reactor. A transmission device is vertically rotatably connected to the base for periodic rotation. A support beam is connected to the transmission device to periodically push and sweep under the actuation of the transmission device. The periodic pushing and sweeping period is an integer multiple of the power frequency period.

[0008] The light intensity distribution of a circular region on the bottom plane of the reactor under different light sources was collected. The grayscale image of the circular region was extracted, and blank data outside the circular region was removed to obtain the data matrix of the circular region.

[0009] Based on image transformation, the data matrix of the circular region is expanded along the radial direction and transformed into a rectangular data matrix through data filling. After expansion, the circumference of the circle becomes the length of the rectangle, the radius of the circle becomes the width of the rectangle, and the data at the center of the circle is copied to become the length of the other side of the rectangle. The data matrix of the circular region becomes a rectangular data matrix, and each column of the rectangular data matrix is ​​the data of the data matrix of the circular region along the radius.

[0010] Principal component analysis is used to perform dimensionality reduction analysis on the rectangular data moments to extract feature points. Multiple SiC ultraviolet solar blind sensor probes are arranged non-uniformly in a linear pattern on the support beam according to the position of the feature points. The non-uniform linear arrangement collects the ultraviolet signal emitted by the reactor in real time and converts it into a voltage signal.

[0011] In the method described, the principal component analysis method divides the support beam into multiple parts based on the length extension direction of the support beam as the radius direction, and performs principal component analysis on each part to obtain feature points.

[0012] In the method described above, based on the physical structure of the reactor, the length extension direction of the support beam is taken as the radius direction and the coordinates are divided into five parts. Principal component analysis is performed on each part to obtain the data features of the corresponding first principal component.

[0013] In the method described, a pseudo-color image is extracted based on the light intensity distribution, and the pseudo-color image is converted into a grayscale image.

[0014] In the method described, the data matrix of the circular region is transformed into a rectangular data matrix and then normalized.

[0015] In the method described, the radius of the data matrix of the circular region is filled with data during image transformation, and the filling value is the average of the data on both sides.

[0016] In the method described, the plurality of SiC ultraviolet solar-blind sensor probes are an array of ultraviolet solar-blind SiC avalanche diode APD sensor probes.

[0017] In the method described, the SiC avalanche diode APD sensor array has a wavelength range of 200nm-400nm and a lower limit of response energy of 1nW / cm². 2 The gain is 10. 5 -10 6 .

[0018] In the method described, the normal to the detection plane of the SiC ultraviolet solar blind sensor probe is parallel to the central axis of the reactor.

[0019] Beneficial effects

[0020] This invention enables online monitoring of dry-type air-core reactors. Compared with traditional electrical monitoring methods, it avoids electromagnetic interference to improve monitoring accuracy, tracks and observes the equipment, detects potential faults early, improves the judgment of poor equipment operating conditions, avoids unnecessary power outages for maintenance and testing, and improves the stability of the power grid and the utilization rate of the equipment. It features high real-time performance and high sensitivity. Utilizing a solar-blind ultraviolet SiC array for discharge location significantly lowers the cost barrier for applying ultraviolet sensing technology.

[0021] The above description is merely an overview of the technical solution of the present invention. In order to make the technical means of the present invention clearer and more understandable, so that those skilled in the art can implement it according to the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more obvious and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0022] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0023] In the attached diagram:

[0024] Figure 1 This is a schematic diagram of optical push-broom positioning of reactor insulation faults according to an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of the structure for optical push-broom localization of reactor insulation faults according to an embodiment of the present invention;

[0026] Figure 3 A simulation model diagram of a reactor for optical push-broom localization of insulation faults according to an embodiment of the present invention;

[0027] Figure 4 A cross-sectional view of the internal structure of a reactor model according to an embodiment of the present invention;

[0028] Figure 5 A light intensity distribution diagram of the bottom plane of a reactor according to an embodiment of the present invention;

[0029] Figure 6 A schematic diagram illustrating the transformation of circular data results into rectangles according to an embodiment of the present invention;

[0030] Figure 7 A schematic diagram of a data conversion process according to an embodiment of the present invention;

[0031] Figure 8 A schematic diagram of a tagged matrix after conversion according to an embodiment of the present invention;

[0032] Figure 9 A schematic diagram of principal component analysis divided into five parts according to an embodiment of the present invention;

[0033] Figure 10 A schematic diagram of the data feature values ​​of the first principal component in the first part of the principal component analysis partitioning according to an embodiment of the present invention;

[0034] Figure 11 A schematic diagram of the data feature values ​​of the first principal component in the second part of the principal component analysis partitioning according to an embodiment of the present invention;

[0035] Figure 12 A schematic diagram of the data feature values ​​of the first principal component in the third part of the principal component analysis partitioning according to an embodiment of the present invention;

[0036] Figure 13 A schematic diagram of the data feature values ​​of the first principal component in the fourth part of the principal component analysis partitioning according to an embodiment of the present invention;

[0037] Figure 14 A schematic diagram of the data feature values ​​of the first principal component in the fifth part of the principal component analysis partitioning according to an embodiment of the present invention.

[0038] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0039] The following will refer to the attached diagram. Figures 1 to 14 Specific embodiments of the invention will be described in more detail below. While specific embodiments of the invention are shown in the accompanying drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0040] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this invention is determined by the appended claims.

[0041] To facilitate understanding of the embodiments of the present invention, the following will provide further explanation and description with reference to the accompanying drawings and several specific embodiments, and the accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0042] like Figures 1 to 2 As shown, the optical push-broom method for locating insulation faults in reactors includes,

[0043] The base 2 is located inside the reactor 1, and the center of the base 2 is collinear with the central axis of the reactor 1.

[0044] The transmission device 3 is vertically rotatably connected to the base 2 to rotate periodically, and the support beam 4 is connected to the transmission device 3 to periodically push and sweep under the actuation of the transmission device 3. The pushing and sweeping period is an integer multiple of the power frequency period.

[0045] The light intensity distribution of a circular region on the bottom plane of the reactor under different light sources was collected. The grayscale image of the circular region was extracted, and blank data outside the circular region was removed to obtain the data matrix of the circular region.

[0046] Based on image transformation, the data matrix of the circular region is expanded along the radial direction and transformed into a rectangular data matrix through data filling. After expansion, the circumference of the circle becomes the length of the rectangle, the radius of the circle becomes the width of the rectangle, and the data at the center of the circle is copied to become the length of the other side of the rectangle. The data matrix of the circular region becomes a rectangular data matrix, and each column of the rectangular data matrix is ​​the data of the data matrix of the circular region along the radius.

[0047] Principal component analysis is used to perform dimensionality reduction analysis on the rectangular data moments to extract feature points. Multiple SiC ultraviolet solar blind sensor probes 5 are arranged non-uniformly in a linear pattern on the support beam 4 according to the position of the feature points. The non-uniform linear arrangement collects the ultraviolet signal emitted by the reactor in real time and converts it into a voltage signal.

[0048] In a preferred embodiment of the method, the principal component analysis method divides the support beam into multiple parts based on the length extension direction of the support beam as the radius direction, and performs principal component analysis on each part to obtain feature points.

[0049] In a preferred embodiment of the method, based on the physical structure of the reactor, the length extension direction of the support beam is taken as the radial direction and the coordinates are divided into five parts. Principal component analysis is performed on each part to obtain the data features of the corresponding first principal component.

[0050] In a preferred embodiment of the method, a pseudo-color image is extracted based on the light intensity distribution, and the pseudo-color image is converted into a grayscale image.

[0051] In a preferred embodiment of the method, the data matrix of the circular region is normalized after being transformed into a rectangular data matrix.

[0052] In a preferred embodiment of the method, data is filled during the radius image transformation of the data matrix of the circular region, and the filling value is the average of the data on both sides.

[0053] In a preferred embodiment of the method, the plurality of SiC ultraviolet solar-blind sensor probes 5 are an array of ultraviolet solar-blind SiC avalanche diode APD sensor probes.

[0054] In a preferred embodiment of the method, the SiC avalanche diode APD sensor array has a wavelength range of 200nm-400nm and a lower limit of response energy of 1nW / cm². 2 The gain is 10. 5 -10 6 .

[0055] In a preferred embodiment of the method, the normal of the detection plane of the SiC ultraviolet solar blind sensor probe 5 is parallel to the central axis of the reactor.

[0056] In one embodiment, the plurality of SiC ultraviolet solar blind sensor probes 5 are gradually distributed more densely in a direction away from the center of the periodic push-broom.

[0057] In one embodiment, the reactor 1 is a dry-type air-core reactor.

[0058] In one embodiment, the ultraviolet sensor probe acquires optical signals in the ultraviolet solar blind band emitted by the internal discharge of reactor 1 in real time.

[0059] In one embodiment, the processing unit is connected to the solar-blind ultraviolet sensor probe and, in response to the voltage signal, synchronously generates ultraviolet light intensity data. The data transmission module, via a Wi-Fi module, enables real-time data transmission, sending the signal to a data platform for subsequent data analysis and diagnostics.

[0060] The ultraviolet sensor probe includes a signal preprocessing unit for preprocessing the acquired ultraviolet signals, an analog-to-digital sampling unit, and an ultraviolet sensor for converting the ultraviolet signals into voltage signals. The signal preprocessing unit includes an amplifier, and the processing unit includes a sampling unit with a sampling rate greater than 100kHz and an amplification factor greater than 100, a detection unit for the envelope peak of the voltage signal, and a calculation unit for obtaining the acoustic intensity of the optical signal through proportional transformation.

[0061] The processing unit includes a digital signal processor, an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA), and includes one or more read-only memories (ROMs), random access memories (RAMs), flash memories or electronically erasable programmable read-only memories (EEPROMs).

[0062] The front surface of the support beam 4 has side-by-side openings so that the ultraviolet sensor array faces the reactor 1; the ultraviolet sensor is welded to the processing unit, and the signal processing unit is connected to the data transmission module via a ribbon cable.

[0063] In one embodiment, the light intensity distribution characteristics of the linear array push-broom at different discharge positions of reactor 1 are obtained through Tracepro simulation; the SiC ultraviolet solar blind sensor probe 5 collects the light intensity data of reactor 1, establishes a database that matches the fault location with the light intensity data, and determines the correspondence between the location of the discharge fault and the sensor data through deep learning and other methods. The ultraviolet signal is detected in the reactor 1 during operation and converted into a voltage signal output. Based on the voltage signal output, deep learning is used to determine whether there is a fault in reactor 1 and to give the diagnosed fault location.

[0064] In one implementation, the method includes,

[0065] For reactor 1, the light intensity distribution characteristics of the linear array push-broom interface at different discharge positions were obtained by Tracepro simulation.

[0066] The light intensity distribution map is normalized and transformed. The data distribution characteristics along the radial direction are analyzed, and the data dimensionality is reduced by principal component analysis to determine the feature points along the radial direction, which are used to place the ultraviolet sensing probe, thus completing the design of the discharge positioning push-broom array device of reactor 1.

[0067] By collecting light intensity data through an ultraviolet sensor, a database is established that matches the fault location with the light intensity data. Through methods such as deep learning, the correspondence between the location of the discharge fault and the sensor data is determined.

[0068] For the dry-type air-core reactor 1 in operation, the push-broom device is used to monitor the ultraviolet solar blindness intensity signal of the reactor 1. Through a deep learning model, it is determined whether there is a fault in the dry-type reactor 1 and the fault location is given.

[0069] Abnormal discharge of reactor 1 can indicate an electrical malfunction in the equipment. Partial discharge detection allows us to identify defects early, address them promptly, and prevent accidents. The discharge process emits ultraviolet light. By using a pushbroom array of ultraviolet solar-blind sensors to locate the partial discharge position, we can predictively detect signs of failure in reactor 1 and take preventative measures to ensure the safe and reliable operation of the equipment.

[0070] The key lies in determining the placement of the sensor on this long rod, as its position directly determines the final detection result. To achieve this, Figure 3 It is a simulation model of the designed reactor. Figure 4 This is a cross-sectional view of the internal structure of the reactor model.

[0071] Light sources were placed in different positions in the designed model, and the coordinates of the light source positions are shown in Table 1:

[0072] Table 1. Locations and coordinates of different partial discharge sources for the reactor.

[0073]

[0074] By simulating the optical path, the light intensity distribution on the bottom plane of the reactor under different light source distributions can be obtained, such as... Figure 5 As shown in the figure, the light intensity distribution on the bottom plane of the reactor varies depending on the position of the light source. Therefore, the location of the fault in the reactor can be determined by observing these different distributions. However, monitoring the entire bottom plane would require too many sensors. Therefore, a linear push-broom array is used as an equivalent replacement. Figure 2 The aforementioned device.

[0075] Determine the placement of the sensor on the bar. Analysis is required. Figure 5 The light intensity distribution is a circular data set. We want to know which locations along the radius of the circle are more important so we can place sensors at those locations. In other words, we need to reduce the dimensionality of the data along the radius. Direct data processing would be difficult. Therefore, we transform the circular data into a rectangular shape, allowing us to use principal component correlation analysis for dimensionality reduction. The specific transformation method is as follows... Figure 6 and Figure 7 As shown. Finally, the light intensity data was converted into... Figure 8 The matrix shown is labeled, and each column contains data collected along the radial direction. Dimensionality reduction of the first 150 rows of this matrix allows us to obtain the optimal sensor coordinates from a global optimization perspective.

[0076] In one embodiment, such as Figure 7 As shown, based on the intensity map, the grayscale image of the circular region is extracted, and the blank data outside the circular region is removed to obtain a set of circular data. Figure 6 As shown, an image transformation algorithm is used to unfold the circle along its radius and, through data padding, convert it into a rectangle. After unfolding, the circumference of the circle becomes the length of the rectangle, and the radius of the circle becomes the width of the rectangle. The data at the center of the circle is copied to become the length of the other side of the rectangle. The radius data also needs to be padded during image transformation, with the padding value being the average of the original data on both sides. After padding, the data matrix of the circular region is transformed into a rectangle, where each column of the rectangle contains the data from the original circle along its radius. Figure 2 Taking a reactor as an example, for a grayscale image of a certain fault, after data transformation and image processing, the result is as follows: Figure 8As shown. After the above image transformation, a new data matrix is ​​obtained, with a width of 150 and a length of 942. Each experiment yields 18 sets of experimental data. Transposing and concatenating these 18 matrices results in a new data matrix with 150 rows and 18*942 columns. Finally, a row is added to the matrix, labeled with numbers as shown in the figure below. Ultimately, a 151*16956 matrix is ​​obtained. Principal component analysis (PCA) can be used to perform dimensionality reduction analysis on the 150 rows of this matrix, extracting key row numbers as data feature points.

[0077] For optical fault detection of reactors, directly using mathematical dimensionality reduction algorithms to optimize sensor position coordinates without considering the reactor's physical structure can easily lead to missed or incorrect feature points. Therefore, considering the reactor's physical structure, we use piecewise principal component analysis (PCA) to perform spatial dimensionality reduction on the matrix to determine the specific sensor placement coordinates. In PCA, the data corresponding to the local peak of the first principal component is selected as feature data. We first divide the coordinates along the radial direction into five parts based on the reactor's physical structure, such as... Figure 9 As shown. Principal component analysis was performed on each part to obtain the eigenvalues ​​of the first principal component, as shown below. Figures 10 to 14 As shown in Table 2, eight local peaks were selected as feature point coordinates.

[0078] Table 2 Feature point coordinates

[0079] serial number <![CDATA[X1]]> <![CDATA[X2]]> <![CDATA[X3]]> <![CDATA[X4]]> <![CDATA[X5]]> <![CDATA[X6]]> <![CDATA[X7]]> <![CDATA[X8]]> Serial Number 6 30 48 103 119 128 139 149 Coordinates (cm) 4 20 32 68.7 79.3 85.3 92.7 99.3

[0080] The effectiveness of feature point coordinates was verified using the KNN algorithm and convolutional neural network algorithm. After selecting the sensor placement location, the model was validated, and the accuracy of partial discharge fault location was diagnosed using three models: fifth-order KNN, seventh-order KNN, and convolutional neural network.

[0081] The steps for the KNN model are as follows:

[0082] 1. Calculate the distance between the test data and each training data point.

[0083] Existing training set samples X t and corresponding label Y t , where X t Contains n data points X t -{x t1 x t2 x t3 x t4 , ..., x tnEach data point is an 8-dimensional vector, representing the light intensity information collected by the 8 sensors. The light intensity value is the average of 942 data points from the corresponding sensors. t The positions of the light sources are divided into 18 categories as described above. This paper uses Euclidean distance as the basis for judging the degree of similarity, and the specific formula is shown in equation (1) below.

[0084]

[0085] Where, dis(x) pi x tj (x) is the test data. tj With training data x pi The Euclidean distance between them, x pil and x til These are the l-th dimension data for this data point.

[0086] 2. Sort according to the increasing distance.

[0087] 3. Select the K points with the smallest distance.

[0088] 4. Determine the frequency of occurrence of the category of the first K points.

[0089] 5. Return the category with the highest frequency among the top K points as the predicted category for the test data.

[0090] The input data consists of light intensity distribution information collected by eight sensors, and the output data is the classification probability. The category with the highest probability is selected as the predicted classification of the input data.

[0091] The accuracy of partial discharge coordinate localization was diagnosed using a dataset with known discharge source coordinates and sensor data. Twenty repeated simulations were conducted, with the light source spatial position randomly adjusted within ±10 cm each time, resulting in 360 sets of light intensity distribution data. The light intensity distribution curves under several characteristic coordinates were used as input feature vectors, with the light source position as the label and output. A K-nearest neighbor classifier and a convolutional neural network were used for training, and cross-validation was employed to calculate the fault location accuracy under different classification models. The recognition accuracy of piecewise PCA dimensionality reduction is shown in Table 3. The results show that fault location based on piecewise PCA dimensionality reduction has high localization accuracy when validated with fifth-order, seventh-order KNN, and convolutional neural network algorithms, indicating that sensor position optimization is effective.

[0092] Table 3

[0093]

[0094] In one embodiment, a dataset with known discharge source coordinates and sensor data is used to diagnose the accuracy of partial discharge coordinate localization. Twenty repeated simulations were conducted, with the light source spatial position randomly adjusted within ±10 cm each time, resulting in 360 sets of light intensity distribution data. The light intensity distribution curves under several characteristic coordinates were used as input feature vectors, and the light source position as the label and output. A K-nearest neighbor classifier was used for training and classification, and a cross-validation algorithm was employed to calculate the fault localization accuracy under different classification models.

[0095] by Figure 1 Taking a reactor model of a specific size as an example, 20 sets of simulation test data were obtained. Each set of data, based on the fault location, included 18 test results, totaling 360 data points. Of these, 270 sets were used for model training and spatial clustering, and 90 sets were used for model validation. The training results of these 270 sets yielded the average response intensity of eight sensors under various fault conditions, forming a set of intensity coordinates, or an 8*842 dimension matrix. The specific process has been detailed above. There are 18 fault types: Upper inner layer fault of Enclosure 1, Lower inner layer fault of Enclosure 1, Upper inner layer fault of Enclosure 2, Lower inner layer fault of Enclosure 2, Upper outer layer fault of Enclosure 2, Lower outer layer fault of Enclosure 2, Upper inner layer fault of Enclosure 3, Lower inner layer fault of Enclosure 3, Upper outer layer fault of Enclosure 3, Lower outer layer fault of Enclosure 3, Upper inner layer fault of Enclosure 4, Lower inner layer fault of Enclosure 4, Upper outer layer fault of Enclosure 4, Lower outer layer fault of Enclosure 4, Upper inner layer fault of Enclosure 5, Lower inner layer fault of Enclosure 5, Upper outer layer fault of Enclosure 5, Lower outer layer fault of Enclosure 5. During training, the fault type and the sensor's 8*842-dimensional response intensity matrix are used as inputs. During validation, the sensor's intensity coordinate matrix is ​​imported, and the Euclidean distance between all fault coordinates in space and the sensor's intensity coordinates is calculated. The fault coordinate closest to this coordinate is considered the actual fault coordinate.

[0096] Although embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the specific embodiments and application fields described above. The specific embodiments described above are merely illustrative and instructive, and not restrictive. Those skilled in the art can make many other forms based on the guidance of this specification and without departing from the scope of protection of the claims of the present invention, and all of these are within the scope of protection of the present invention.

Claims

1. A method for optical push-broom localization of insulation faults in reactors, characterized in that, It includes the following steps, The base is located inside the reactor and the center of the base is collinear with the central axis of the reactor. The transmission device is vertically rotatably connected to the base to rotate periodically. The support beam is connected to the transmission device to periodically push and sweep under the actuation of the transmission device. The pushing and sweeping period is an integer multiple of the power frequency period. The light intensity distribution of a circular region on the bottom plane of the reactor under different light sources was collected. The grayscale image of the circular region was extracted, and the blank data outside the circular region was removed to obtain the data matrix of the circular region. Based on image transformation, the data matrix of the circular region is expanded along the radial direction and transformed into a rectangular data matrix through data filling. After expansion, the circumference of the circle becomes the length of the rectangle, the radius of the circle becomes the width of the rectangle, and the data at the center of the circle is copied to become the length of the other side of the rectangle. The data matrix of the circular region becomes a rectangular data matrix, and each column of the rectangular data matrix is ​​the data of the data matrix of the circular region along the radius. Principal component analysis (PCA) is used to perform dimensionality reduction analysis on the rectangular data moments to extract feature points. Multiple SiC ultraviolet (UV) solar-blind sensor probes are linearly and non-uniformly arranged on the supporting beam according to the positions of these feature points. This non-uniform linear arrangement acquires the UV signal emitted by the reactor in real time and converts it into a voltage signal. The multiple SiC UV solar-blind sensor probes gradually increase in density away from the center of the periodic sweep. These probes acquire the optical signal in the UV solar-blind frequency band emitted by the internal discharge of the reactor in real time. These multiple SiC UV solar-blind sensor probes are a UV solar-blind SiC avalanche diode (APD) sensor array. The wavelength range of the SiC avalanche diode (APD) sensor array includes 200nm-400nm, and its lower limit of response energy is 1nW / cm². 2 The gain is 10. 5 -10 6 .

2. The method according to claim 1, characterized in that: The principal component analysis method divides the support beam into multiple parts based on the length extension direction of the support beam as the radius direction, and performs principal component analysis on each part to obtain feature points.

3. The method according to claim 2, characterized in that: Based on the physical structure of the reactor, the length extension direction of the support beam is taken as the radius direction and the coordinates are divided into five parts. Principal component analysis is performed on each part to obtain the data features of the corresponding first principal component.

4. The method according to claim 1, characterized in that: A pseudo-color image is extracted based on the light intensity distribution, and the pseudo-color image is converted into a grayscale image.

5. The method according to claim 1, characterized in that: After the data matrix of the circular region is transformed into a rectangular data matrix, it is normalized.

6. The method according to claim 1, characterized in that: When the radius of the data matrix of the circular region is transformed, data is filled with the mean of the data on both sides.

7. The method according to claim 1, characterized in that: The normal to the detection plane of the SiC ultraviolet solar blind sensor probe is parallel to the central axis of the reactor.

Images