Method and equipment for comprehensively evaluating state of valve hall equipment in combination with multispectrum and data cube

Abstract

The invention relates to a valve hall equipment state comprehensive evaluation method combining multispectrum and a data cube. The method comprises the following steps: constructing a valve hall three-dimensional grid space scene; constructing a multispectral image data set; constructing a valve hall equipment multispectral index analysis system; integrating into a data cube; obtaining subjective and objective comprehensive weights of valve hall equipment state evaluation; carrying out quantitative sorting on the equipment states, and giving an evaluation result; the state of the valve hall equipment is comprehensively evaluated, and potential defects and problems are identified. According to the method, a complete target equipment evaluation index system is constructed, various technical indexes for objectively quantifying the equipment state are given, subjective and objective quantification is performed based on AHP and entropy weight, the defect of single evaluation is avoided, and accurate evaluation weight is obtained based on the D-S evidence theory; the method improves the efficiency and precision of operation and maintenance, reduces the operation and maintenance cost, powerfully supports the formulation and implementation of a predictive maintenance strategy, and improves the safety and reliability of equipment operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of valve hall equipment defect evaluation, and in particular to a valve hall equipment state comprehensive evaluation method and device combining multispectrum and data cubes. BACKGROUND

[0002] In the field of electric power industry, valve hall equipment is a key component of an extra-high voltage converter station, and its operating condition is directly related to the safety and stability of the power system. However, the traditional valve hall equipment defect diagnosis method mostly relies on manual experience inspection, and mainly has the following problems:

[0003] Firstly, when performing collaborative diagnosis on valve hall equipment, the multispectrum information composed of infrared light, ultraviolet light and visible light is not fully utilized, and the correlation analysis among the three is lacking, resulting in insufficient evaluation indexes and insufficient dimensions, which makes it difficult to comprehensively and accurately reflect the potential defects and overall state of the equipment.

[0004] Secondly, the multispectrum information generated by equipment inspection is complex and diverse, and lacks reasonable and efficient data organization, feature extraction and analysis methods, resulting in low data screening efficiency. The rich data contained therein cannot be fully mined and utilized, which seriously affects the accuracy and timeliness of the state evaluation.

[0005] Thirdly, the existing evaluation method cannot effectively fuse the multisource, heterogeneous and high-dimensional feature information extracted from the multispectrum data cube, lacks a scientific and reasonable feature weight distribution mechanism, and lacks the ability to objectively quantify and sort the equipment state. SUMMARY

[0006] To solve the problems of incomplete valve hall equipment state evaluation, low data organization and analysis efficiency, and unreasonable feature weight distribution in the evaluation process, the primary purpose of the present application is to provide a complete target equipment evaluation index system, and to give multiple technical indexes for objectively quantifying the equipment state, thereby avoiding the shortcomings of single evaluation and realizing the comprehensive evaluation method of valve hall equipment state combining accurate evaluation weight and multispectrum and data cubes.

[0007] To achieve the above purpose, the present application adopts the following technical solution: a comprehensive evaluation method of valve hall equipment state combining multispectrum and data cubes, which comprises the following sequential steps:

[0008] (1) constructing a three-dimensional grid space scene of the valve hall to obtain the spatial identifier of the equipment;

[0009] (2) constructing a multispectrum image data set for evaluating the target equipment with acquisition time;

[0010] (3) Construct a valve hall equipment multispectral index analysis system identified by a multispectral image dataset;

[0011] (4) Integrate the four-dimensional data of spatial coordinate identifiers, equipment attributes, equipment multispectral indexes, and collection times into a data cube;

[0012] (5) Perform OLAP operations on the data cube to extract quantitative features and obtain subjective and objective comprehensive weights for valve hall equipment state evaluation;

[0013] (6) Quantitatively sort the equipment state based on the obtained subjective and objective comprehensive weights for valve hall equipment state evaluation, and give the corresponding evaluation results;

[0014] (7) According to the evaluation results, a comprehensive evaluation is carried out on the state of the valve hall equipment, and potential defects and problems are identified.

[0015] Step (1) specifically refers to: according to the actual physical space structure of the valve hall, the continuous space inside the valve hall is constructed into a regular three-dimensional grid unit, and a grid framework with spatial position as the index is built, which provides a unified spatial reference for multispectral data, and each three-dimensional grid unit is assigned a unique spatial identifier space. The spatial identifier space is composed of a voxel grid coordinate and a station field, the voxel grid coordinate is represented by three-dimensional coordinates (x, y, z), which correspond to the coordinates in X, Y, and Z directions respectively, thereby obtaining the spatial coordinates of the equipment, and station is the station number corresponding to the valve hall.

[0016] Step (2) specifically includes the following steps in order:

[0017] (2a) Arrange multiple multispectral cameras to collect equipment multispectral images in a three-dimensional grid scene, and record the time of each collection;

[0018] (2b) Perform image filtering using a non-subsampled contourlet transform formed by combining a non-subsampled pyramid transform and a non-subsampled directional filter bank;

[0019] (2c) Segment the multispectral image by an improved U-Net network to identify the key components of the valve hall equipment, wherein the key components refer to the core components important to the operation of the high-voltage direct current transmission system in the valve hall; and the improved U-Net network is a CBAM module added between the double convolution and pooling layers of each downsampling block;

[0020] (2d) Add equipment asset numbers and spatial identifiers as labels to the segmented images in sequence.

[0021] Step (3) specifically includes the following steps in order:

[0022] (3a) Constructing the infrared image evaluation index set of the target device: the infrared image evaluation index set includes maximum temperature T max , minimum temperature T min , hotspot maximum relative temperature rise T improve , temperature non-uniformity T uneven , history average same period temperature rise T imp , and hotspot area ratio S ratio ; let the area of the region of interest segmented from the infrared image be S, and define the temperature non-uniformity T uneven as follows:

[0023]

[0024] In the formula, T avg is the average temperature;

[0025] Let the ambient temperature be T env , and the hotspot maximum temperature be T max_rel , then the hotspot maximum relative temperature rise T improve is:

[0026] T improve = T max_rel -T env ;

[0027] Let the history same period average be T his_avg , then the history average same period temperature rise T imp is:

[0028] T imp = T avg -T his_avg ;

[0029] Let the extracted hotspot area be S hot , then the hotspot area ratio S ratio is:

[0030]

[0031] (3b) Constructing the ultraviolet image evaluation index set of the target device: the ultraviolet image evaluation index set includes discharge area area ratio S dis_ratio , discharge hotspot number ratio M hot_ratio , maximum photon density per unit area ρ max , average photon density per unit area ρ avg , and maximum discharge intensity E max Five indexes; let the area of the region of interest segmented from the ultraviolet image be S1, and the total discharge area area be S discharge , then the discharge area area ratio S dis_ratio is:

[0032]

[0033] (3c) constructing a visible light image evaluation index set of the target device, the visible light image evaluation index set including a rust area proportion S rust , a bolt loosening angle A bolt , and a crack area proportion S crack ;

[0034] (3d) the infrared image evaluation index set, the ultraviolet image evaluation index set, and the visible light image evaluation index set jointly constitute a valve hall device multispectral index analysis system.

[0035] Step (4) specifically includes the following sequential steps:

[0036] (4a) first constructing a four-dimensional data cube SpectralDataCube:

[0037] SpectralDataCube={dimension,hierarchy,measure};

[0038] dimension={device,space,multispectral,time};

[0039] hierarchy={hierarchy1,hierarchy2,…,hierarchy n};

[0040] measure={measure1,measure2,…,measure n};

[0041] In the formula, dimension is a multispectral data cube dimension expression, and the multispectral cube is set to have four dimensions, so as to form a four-dimensional tensor data space: hierarchy represents the relationship between the concepts to which the data correspond; measure is a classification characteristic value corresponding to the multispectral image data cube, and n is the total number of characteristics; device represents the device dimension, space represents the spatial grid position dimension, multispectral represents the multispectral dimension, and time represents the time dimension;

[0042] (4b) setting the specific attributes involved in the device dimension, which are composed of site numbers, device names, and device numbers; the space is composed of grid cell coordinates and site numbers; the time is determined by the multispectral image sampling time interval, and forms a one-dimensional sampling time sequence; the multispectral dimension is composed of visible light, infrared light, and ultraviolet light.

[0043] Step (5) specifically includes the following sequential steps:

[0044] (5a) OLAP operation is carried out on the data cube to obtain results accumulated in different dimensions, and the OLAP operation includes drilling, slicing and dicing;

[0045] (5b) Obtain the subjective weight W of the target device based on AHP a =(w a1 ,w a2 ,…w an ), wherein n is the total number of indexes;

[0046] (5c) Obtain the objective weight of the target device: use the improved entropy weight method to quantify the results accumulated in different dimensions obtained through the data cube, evaluate the relative importance of each key component of the valve hall device, and obtain the objective weight value; the improved entropy weight method is specifically:

[0047] For two variables L x =(l 1x ,l 2x ,…,l Nx ) and L y =(l 1y ,l 2y ,…,l Ny ) containing N dimensions, arrange L x and L y in ascending order to obtain the sorting value vectors and The rank correlation coefficient r xy is:

[0048]

[0049] In the formula, and respectively represent the mean value of the sorting value vectors of , and respectively represent the i-th element of , the i-th element of , and N is the dimension of ;

[0050] After obtaining the rank correlation coefficient r xy , the objective weight W b =(ω b1 ,ω b2 ,…ω bn ) can be obtained, which satisfies:

[0051]

[0052] In the formula, r jy is the rank correlation coefficient of the j-th index in r xy , and M represents the number of indexes in rjy w ej jth element weight obtained by the entropy weight method before improvement, W b jth element of w

[0053] (5d) Obtain the subjective and objective comprehensive weight by combining AHP with the improved entropy weight method: assuming that there are m evaluation indexes, on the basis of obtaining the subjective and objective weights respectively, according to the D-S evidence theory, the subjective and objective comprehensive weight w j is:

[0054]

[0055] In the formula, w aj , w bj are the subjective weight and the objective weight of the jth index respectively, is the combination symbol in the D-S evidence theory.

[0056] Step (6) specifically includes the following sequential steps:

[0057] (6a) Divide the selected indexes into the target device index health level percentage system, and the higher the score, the better the device health degree;

[0058] (6b) Obtain the score of the device to be evaluated according to the index health level: assuming that the scores of the indexes are s1, s2, …, s n , and the subjective and objective comprehensive weights are w1, w2, …, w n , then the total health degree score H e of the device to be evaluated is:

[0059] H e =∑ i (w i s i );

[0060] In the formula, i∈[1,n], n represents the total number of indexes, w i is the subjective and objective comprehensive weight of the ith index, and s i is the score of the ith index;

[0061] (6c) Select the device health degree evaluation, and present the scores of the indexes in the form of a radar chart: correspond each index name to an endpoint of the radar chart, mark the values on the corresponding axes according to the scores s1, s2, …, s n , and connect the marking points in sequence according to the index order to form the distribution form of the radar chart;

[0062] (6d) Place the calculated total health degree score H e in the concentric circular ring chart on the right side of the radar chart, and the concentric circular ring chart includes four rings, He The scores in [0, 59], (60, 79], (80, 89], [90, 100] are poor, pass, good, and excellent, respectively. e The evaluation result is obtained by hitting the annular ring.

[0063] Step (2b) specifically comprises the following steps in sequence:

[0064] (2b1) The image multi-scale analysis is realized by cascaded pyramid decomposition, and first, the non-subsampled pyramid transformation is used for multi-scale decomposition of the source image;

[0065] (2b2) The non-subsampled directional filter bank is used for further multi-directional decomposition of the high-frequency component, so as to obtain the sub-band image under different scales and different directions;

[0066] (2b3) The obtained sub-band image is set to a threshold for processing to filter out noise, and a reasonable threshold is set to remove weak signals and noise in the image;

[0067] (2b4) The image after the non-subsampled contourlet transformation is reconstructed to obtain the denoised multi-spectral image, improve the signal-to-noise ratio and definition of the image, and provide high-quality image data for subsequent feature extraction and analysis.

[0068] Another purpose of the present application is to provide an electronic device comprising:

[0069] A processor; and

[0070] A memory, in which computer program instructions are stored, the computer program instructions, when executed by the processor, cause the processor to perform the valve hall equipment state comprehensive evaluation method combined with multi-spectrum and data cube as described above.

[0071] The present application also provides a computer readable storage medium having computer program instructions stored thereon, the computer program instructions, when executed by a processor, cause the processor to perform the valve hall equipment state comprehensive evaluation method combined with multi-spectrum and data cube as described above.

[0072] From the above technical solutions, the beneficial effects of the present application are: first, combined with multispectral imaging and data cube technology, the multispectral image data can be efficiently integrated, overcoming the problems of traditional data processing procedures such as complicated steps, low query efficiency, and data islandization; second, a complete target device evaluation index system is constructed, giving a variety of technical indexes for objective quantification of device status, and based on AHP and entropy weight for subjective and objective quantification, avoiding the shortcomings of single evaluation, and based on D-S evidence theory, accurate evaluation weight is obtained; third, combined with spatial gridding processing and data cube storage structure, the fault position can be effectively obtained, and the untimely situation caused by manual inspection is alleviated, the present application provides an integrated solution for valve hall device state monitoring and evaluation, improves the efficiency and accuracy of operation and maintenance, reduces the operation and maintenance cost, with the provision of quantitative and intuitive state evaluation results and fault position information, it strongly supports the formulation and implementation of predictive maintenance strategy, and improves the safety and reliability of equipment operation. BRIEF DESCRIPTION OF DRAWINGS

[0073] Figure 1 is a flow chart of the method of the present application;

[0074] Figure 2 is a four-dimensional data cube and its slice operation schematic diagram;

[0075] Figure 3 is a valve hall device multispectral index analysis system schematic diagram. DETAILED DESCRIPTION

[0076] As shown in Figure 1 , a valve hall device state comprehensive evaluation method combining multispectral and data cube, the method comprises the following sequential steps:

[0077] (1) constructing a valve hall three-dimensional gridding space scene to obtain the spatial identifier of the device;

[0078] (2) constructing a multispectral image dataset for evaluating the target device with acquisition time;

[0079] (3) constructing a valve hall device multispectral index analysis system identified by the multispectral image dataset; the construction of this system is quite critical for accurate assessment of the valve hall device state, and the device index is an intuitive basis for reflecting the device state, and its criticality is obvious. In actual valve hall device state evaluation work, if there is a lack of complex and comprehensive device index, the accurate evaluation of the device state may result in a situation of one-sided evaluation.

[0080] (4) The spatial coordinate identifier, device attribute, device multispectral index and four-dimensional data of acquisition time are integrated into a data cube; the data cube technology is used to organize and manage the multispectral data in a multidimensional space, and such data management form can effectively analyze, mine and process the data, and provide more information support for comprehensive evaluation of the valve hall equipment state.

[0081] (5) The OLAP operation is performed on the data cube to extract quantitative features and obtain the subjective and objective comprehensive weight of the valve hall equipment state evaluation;

[0082] (6) The valve hall equipment state is quantitatively sorted according to the subjective and objective comprehensive weight of the valve hall equipment state evaluation, and the corresponding evaluation result is given;

[0083] (7) According to the evaluation result, the comprehensive evaluation of the valve hall equipment state is carried out, and the potential defects and problems are identified.

[0084] Step (1) is specifically: according to the actual physical space structure of the valve hall, the continuous space inside the valve hall is constructed into a regular three-dimensional grid unit, and a grid framework with spatial position as index is built, which provides a unified spatial reference for multispectral data, and each three-dimensional grid unit is assigned a unique spatial identifier space. The spatial identifier space is composed of voxel grid coordinates and station field, the voxel grid coordinates are represented by three-dimensional coordinates (x, y, z), which correspond to the coordinates in X, Y and Z directions respectively, thereby obtaining the spatial coordinates of the equipment, and station is the station number corresponding to the valve hall.

[0085] Step (2) specifically includes the following steps in order:

[0086] (2a) A plurality of multispectral cameras are arranged in the three-dimensional grid scene to collect multispectral images of the equipment, and the time of each collection is recorded;

[0087] (2b) The non-subsampled contourlet transform formed by combining the non-subsampled pyramid transform and the non-subsampled directional filter bank is used for image filtering;

[0088] (2c) The multispectral image is segmented by the improved U-Net network to identify the key components of the valve hall equipment, wherein the key components refer to the core components important to the operation of the high-voltage direct current transmission system in the valve hall; the improved U-Net network is a CBAM module added between the double convolution and pooling layer of each down-sampling block;

[0089] (2d) The segmented image is sequentially labeled with device asset number and spatial identifier.

[0090] Step (3) specifically includes the following steps in order:

[0091] (3a) Constructing the infrared image evaluation index set of the target device: the infrared image evaluation index set includes the maximum temperature T max , the minimum temperature T min , the maximum relative temperature rise T improve of the hot spot, the temperature non-uniformity T uneven , the history average same period temperature rise T imp , and the hot spot area ratio S ratio ; assuming that the area of the region of interest segmented from the infrared image is S, the temperature non-uniformity T uneven is defined as:

[0092]

[0093] In the formula, T avg is the average temperature;

[0094] Assuming that the ambient temperature is T env , the maximum temperature of the hot spot is T max_rel , the maximum relative temperature rise T improve of the hot spot is:

[0095] T improve = T max_rel -T env ;

[0096] Assuming that the history average same period temperature is T his_avg , the history average same period temperature rise T imp is:

[0097] T imp = T avg -T his_avg ;

[0098] Assuming that the extracted hot spot area is S hot , the hot spot area ratio S ratio is:

[0099]

[0100] By calculating the above six indicators involved, the uniformity of temperature distribution in the infrared image and the temperature change condition of the hot spot area can be quantitatively evaluated, thereby analyzing the thermal state of the target device in the valve hall.

[0101] (3b) Constructing the ultraviolet image evaluation index set of the target device: the ultraviolet image evaluation index set includes the discharge area ratio S dis_ratio , the discharge hot spot number ratio M hot_ratio , the maximum photon density per unit area ρ max , the average photon density per unit area ρ avg , and the maximum discharge intensity Emax Five indicators; the area of the region of interest segmented from the ultraviolet image is S1, the total area of the discharge region is S discharge , and the area ratio of the discharge region is S dis_ratio .

[0102]

[0103] The content fed back by the ultraviolet image evaluation index set is mainly the discharge information presented by the surface of the target device. By extracting the discharge-related indicators, the relevant characteristics of the device can be effectively reflected, and the correlation between the device insulation state and the degree of deterioration can be established. The main use of ultraviolet imaging technology is to detect corona discharge and surface partial discharge, and these phenomena are early manifestations of device insulation defects or aging.

[0104] (3c) Constructing a visible light image evaluation index set of the target device, the visible light image evaluation index set including a rust area ratio S rust , a bolt loosening angle A bolt , and a crack area ratio S crack ; the visible light image can present many common information existing on the surface of the target device, can quantify the image quality itself, and can also analyze the typical defects such as cracks and deformation on the surface of the device by means of visual features.

[0105] (3d) The infrared image evaluation index set, the ultraviolet image evaluation index set, and the visible light image evaluation index set together constitute a valve hall device multispectral index analysis system. By collecting these three indicators, the device deformation condition can be effectively obtained, and the progressive deterioration process of the mechanical structure of the device can be further accurately captured.

[0106] Step (4) specifically includes the following sequential steps:

[0107] (4a) First, construct a four-dimensional data cube SpectralDataCube:

[0108] SpectralDataCube = {dimension, hierarchy, measure};

[0109] dimension = {device, space, multispectral, time};

[0110] hierarchy = {hierarchy1, hierarchy2, …, hierarchy n};

[0111] measure = {measure1, measure2, …, measuren};

[0112] dimension is the multi-spectral data cube dimension expression, setting the multi-spectral cube has 4 dimensions, which can form a four-dimensional tensor data space: hierarchy represents the relationship between the concepts corresponding to the data; measure is the classification feature value corresponding to the multi-spectral image data cube, n is the total number of features; device represents the device dimension, space represents the spatial grid position dimension, multispectral represents the multi-spectral dimension, and time represents the time dimension;

[0113] (4b) Set the specific attributes involved in the device dimension, which is composed of site number, device name and device number; space is composed of grid cell coordinates and site number; time is determined by the multi-spectral image sampling time interval, and forms a one-dimensional sampling time sequence; the multispectral dimension is composed of visible light, infrared light and ultraviolet light.

[0114] Step (5) specifically includes the following steps in sequence:

[0115] (5a) OLAP operation is carried out on the data cube to obtain the results accumulated in different dimensions, and the OLAP operation includes drilling, slicing and dicing; drilling operation can obtain multi-spectral image data of fixed dimension, slicing operation on the data cube can obtain multi-spectral image data of a specific point, and dicing operation on the cube can obtain a section of interval, a group of interval data, which can store multi-spectral image data in an efficient way by means of these operations.

[0116] (5b) Obtain the target device subjective weight W based on AHP a =(w a1 ,w a2 ,…w an ), where n is the total number of indexes;

[0117] (5c) Obtain the objective weight of the target device: use the improved entropy weight method to quantify the results accumulated in different dimensions obtained through the data cube, evaluate the relative importance of each key component of the valve hall device, and obtain the objective weight value; the improved entropy weight method specifically refers to:

[0118] For two variables L x =(l 1x ,l 2x ,…,l Nx ) and L y =(l 1y ,l 2y ,…,l Ny ) containing N dimensions, L x , Ly ascending order, obtaining a ranking value vector with rank correlation coefficient r xy is:

[0119]

[0120] wherein, with respectively represent the mean of the ranking value vector, respectively represent the i-th element of the ranking value vector, the i-th element of the ranking value vector, and N is the dimension of the ranking value vector;

[0121] After obtaining the rank correlation coefficient r xy , the objective weight W b =(ω b1 ,ω b2 ,…ω bn ) can be obtained, which satisfies:

[0122]

[0123] wherein, r jy is the rank correlation coefficient of the j-th index in r xy , M represents the range of r jy , w ej is the j-th element weight obtained by the entropy weight method before improvement, is the j-th element of W b ;

[0124] (5d) Obtain the subjective and objective comprehensive weight by combining AHP and improved entropy weight method: assuming that there are m evaluation indexes, on the basis of obtaining the subjective and objective weights respectively, according to the D-S evidence theory, the subjective and objective comprehensive weight w j is:

[0125]

[0126] wherein, w aj , w bj are the subjective weight and the objective weight of the j-th index respectively, is the combination symbol in the D-S evidence theory.

[0127] Step (6) specifically includes the following sequential steps:

[0128] (6a) Divide the selected indicators into a percentage system for the health level of the target equipment indicators. The higher the score, the better the health level of the equipment. Assume that the scores of each indicator are all out of 100 and they are divided into ten levels, that is, with 10 as the dividing interval. [0,10] is the lowest level and [90,100] is the highest level. The better the level, the higher the quality and the lower the degree of deterioration.

[0129] (6b) The equipment to be evaluated is scored according to the health level of the indicators: Assume the scores of each indicator are s1, s2, ..., s n The subjective and objective weights are w1, w2, ..., w n The overall health score H of the equipment to be evaluated is... e :

[0130] H e =∑ i (w i s i );

[0131] In the formula, i∈[1,n], n represents the total number of indicators, w i Let s be the combined subjective and objective weights of the i-th indicator. i The score for the i-th indicator;

[0132] (6c) Select equipment health assessment and present the scores of each indicator in the form of a radar chart: map each indicator name to one endpoint of the radar chart, and assign each indicator score to an endpoint s1, s2, ..., s2. n Mark the values ​​on the corresponding axes, and connect the marked points in sequence according to the index to form the distribution pattern of the radar chart.

[0133] (6d) Calculate the total health score H e The concentric donut chart placed to the right of the radar chart includes four rings, H e Scores in [0,59], (60,79], (80,89], and [90,100] represent the range, pass, good, and excellent, respectively. The calculated H... e Hit the ring to obtain the evaluation result.

[0134] Step (2b) specifically includes the following steps in sequence:

[0135] (2b1) Multi-scale image analysis is achieved through cascaded pyramid decomposition. First, the source image is decomposed into multiple scales using non-subsampled pyramid transformation.

[0136] (2b2) ​​The high-frequency components are further decomposed in multiple directions using a non-subsampled directional filter bank to obtain sub-band images at different scales and in different directions;

[0137] (2b3) The obtained sub-band image is processed by setting a threshold to filter out noise. A reasonable threshold is set to remove weak signals and noise in the image.

[0138] (2b4) The image after non-subsampled contour wave transformation is reconstructed to obtain a denoised multispectral image, which improves the signal-to-noise ratio and clarity of the image and provides high-quality image data for subsequent feature extraction and analysis.

[0139] Figure 2 This involves a four-dimensional data cube and its slicing operation. First, a four-dimensional data cube is shown, composed of space, device, multispectral, and time dimensions. Faces ABCD represent a cross-section of this four-dimensional space. The multispectral dimension is divided into three layers: ultraviolet (IR), visible light (VIS), and infrared (UV). Performing a time-related OLAP slicing operation on this four-dimensional data cube—that is, fixing a specific point in time—results in a three-dimensional data cube. Figure 2 The three-dimensional data cube in the image is a concrete description of the data, reflecting the specific data situation at the same point in time. Each block of the cube stores a dataset of device image indicators.

[0140] Figure 3 This is a schematic diagram of the multispectral index analysis system for valve hall equipment. The system consists of four layers. The first three layers are, in order, equipment image data, equipment image multispectral target layer, and corresponding criterion layer. The fourth layer is the specific index layer, which reflects the indicators extracted from the image source layer. With the help of these indicators, the characteristics of the equipment can be presented more objectively and comprehensively.

[0141] In summary, by combining multispectral imaging and data cube technology, multispectral image data can be efficiently integrated, overcoming the problems of complex steps, low query efficiency, and data silos in traditional data processing workflows. A complete evaluation index system for target equipment is constructed, providing multiple technical indicators for objectively quantifying equipment status. Subjective and objective quantification is performed based on AHP and entropy weights, avoiding the shortcomings of single evaluations. Furthermore, accurate evaluation weights are obtained based on DS evidence theory. Combined with spatial gridding processing and a data cube storage structure, fault locations can be effectively obtained, alleviating the untimely nature of manual inspections. This invention provides an integrated solution for valve hall equipment status monitoring and evaluation, improving the efficiency and accuracy of operation and maintenance work, reducing operation and maintenance costs, and providing quantitative and intuitive status assessment results and fault location information. It strongly supports the formulation and implementation of predictive maintenance strategies, improving the safety and reliability of equipment operation.

[0142] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A comprehensive evaluation method for the condition of valve hall equipment combining multispectral and data cube approaches, characterized in that: The method includes the following steps in sequence: (1) Construct a three-dimensional meshed spatial scene of the valve hall to obtain the spatial identifiers of the equipment; (2) Construct a multispectral image dataset with acquisition time for evaluating the target device; (3) Construct a multispectral index analysis system for valve hall equipment identified by multispectral image datasets; (4) Integrate spatial coordinate identifiers, equipment attributes, equipment multispectral indicators and four-dimensional data of acquisition time into a data cube; (5) Perform OLAP operation on the data cube to extract quantitative features and obtain the subjective and objective comprehensive weights for the valve hall equipment status evaluation. (6) The equipment status is quantitatively ranked by the subjective and objective comprehensive weights of the obtained valve hall equipment status evaluation, and the corresponding evaluation results are given. (7) Based on the evaluation results, a comprehensive evaluation of the condition of the valve hall equipment is carried out to identify potential defects and problems.

2. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube analysis as described in claim 1, characterized in that: Step (1) specifically refers to: based on the actual physical spatial structure of the valve hall, constructing the continuous space inside the valve hall into regular three-dimensional grid units, building a grid framework with spatial location as the index, thereby providing a unified spatial reference for multispectral data, and assigning a unique spatial identifier space to each three-dimensional grid unit. The spatial identifier space consists of voxel grid coordinates and the station field. The voxel grid coordinates are represented by three-dimensional coordinates (x, y, z), which correspond to the coordinates in the X, Y, and Z directions, respectively, thereby obtaining the spatial coordinates of the equipment. The station is the station number corresponding to the valve hall.

3. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube analysis as described in claim 1, characterized in that: Step (2) specifically includes the following steps in sequence: (2a) Arrange multiple multispectral cameras in a 3D meshed scene to acquire multispectral images and record the time of each acquisition. (2b) Image filtering is performed by combining non-subsampled pyramid transform and non-subsampled directional filter bank to form non-subsampled contour wave transform; (2c) The multispectral image is segmented by an improved U-Net network to identify the key components of the valve hall equipment. The key components refer to the core components in the valve hall that are important for the operation of the high voltage DC transmission system. The improved U-Net network adds a CBAM module between the double convolution and pooling layers of each downsampling block. (2d) Add the device asset number and spatial identifier as labels to the segmented image in sequence.

4. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube as described in claim 1, characterized in that: Step (3) specifically includes the following steps in sequence: (3a) Constructing the infrared image evaluation index set for the target device: The infrared image evaluation index set includes the maximum temperature T max Minimum temperature T min The relative temperature rise of hotspots (T) improve Temperature nonuniformity T uneven Historical average year-on-year temperature rise T imp and hotspot area ratio S ratio Let S be the area of ​​the region of interest segmented from the infrared image, and define the temperature non-uniformity T. uneven for: In the formula, T avg It is the average temperature; Let the ambient temperature be T. env The maximum temperature of the hotspot is T. max_rel The extreme value of the hot spot is relative to the temperature rise T. improve for: T improve =T max_rel -T env ; Let the historical average for the same period be T. his_avg The historical average year-on-year temperature rise T imp for: T imp =T avg -T his_avg ; Let the extracted hotspot area be S. hot The area of ​​hotspots is S ratio for: (3b) Constructing a set of ultraviolet image evaluation indicators for the target device: The ultraviolet image evaluation indicator set includes the discharge region area ratio S dis_ratio The proportion of discharge hotspots M hot_ratio Maximum photon density per unit area ρ max Average photon density per unit area ρ avg and maximum discharge intensity E max Five metrics; let the area of ​​the region of interest segmented from the ultraviolet image be S1, and the total area of ​​the discharge region be S... discharge Then the area of ​​the discharge region accounts for S dis_ratio for: (3c) Construct a visible light image evaluation index set for the target equipment. The visible light image evaluation index set includes the corrosion area percentage S. rust Bolt loosening angle A bolt and the proportion of crack area S crack ; (3d) The infrared image evaluation index set, the ultraviolet image evaluation index set, and the visible light image evaluation index set together constitute the multispectral index analysis system for valve hall equipment.

5. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube analysis according to claim 1, characterized in that: Step (4) specifically includes the following steps in sequence: (4a) First, construct the four-dimensional data cube SpectralDataCube: SpectralDataCube={dimension,hierarchy,measure}; dimension={device,space,multispectral,time}; hierarchy={hierarchy1,hierarchy2,…,hierarchy n }; measure={measure1,measure2,…,measure n }; In the formula, dimension represents the dimension of the multispectral data cube. The multispectral cube is set to have 4 dimensions, which can form a four-dimensional tensor data space: hierarchy represents the relationship between the concepts corresponding to the data; measure is the classification feature value corresponding to the multispectral image data cube, n is the total number of features; device represents the device dimension, space represents the spatial gridded location dimension, multispectral represents the multispectral dimension, and time represents the time dimension. (4b) Define the specific attributes involved in the device dimension, which consists of the site number, device name, and device number concatenated together; space consists of the grid cell coordinates and the site number; time is determined by the multispectral image sampling time interval and constitutes a one-dimensional sampling time series; the multispectral dimension consists of visible light, infrared light, and ultraviolet light.

6. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube as described in claim 1, characterized in that: Step (5) specifically includes the following steps in sequence: (5a) Perform OLAP operations on the data cube to obtain the results of accumulation in different dimensions. OLAP operations include drill-down, slicing and dicing. (5b) Obtain the subjective weight W of the target device based on AHP a =(w a1 ,w a2 ,…w an ), where n is the total number of indicators; (5c) Obtaining the objective weights of the target equipment: The improved entropy weight method is used to quantify the results of different dimension aggregations obtained through the data cube, evaluate the relative importance of each key component of the valve hall equipment, and obtain objective weight values; the improved entropy weight method specifically refers to: For two variables L, each with N dimensions x =(l 1x ,l 2x ,…,l Nx L y =(l 1y ,l 2y ,…,l Ny ), will L x L y Sort in ascending order to obtain a sorted value vector. and Rank correlation coefficient r xy for: In the formula, and They represent The mean of the sorted value vector. They represent The i-th element The i-th element, N is dimensionality; After obtaining the rank correlation coefficient r xy Then the objective weight W can be obtained. b =(ω b1 ,ω b2 ,…ω bn )satisfy: In the formula, r jy For r xy The rank correlation coefficient of the j-th indicator in the equation, where M represents r. jy The range; w ej The weight of the j-th element obtained by the entropy weight method before the improvement. For W b The j-th element; (5d) Obtaining the combined subjective and objective weights by combining AHP with the improved entropy weight method: Assuming there are m evaluation indicators, based on the separate calculation of subjective and objective weights, the combined subjective and objective weight w can be obtained according to the DS evidence theory. j for: In the formula, w aj w bj These are the subjective weight and objective weight of the j-th indicator, respectively. This is a composite symbol in the DS evidence theory.

7. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube as described in claim 1, characterized in that: Step (6) specifically includes the following steps in sequence: (6a) Divide the selected indicators into a percentage system for the health level of the target equipment indicators. The higher the score, the better the health level of the equipment. (6b) The equipment to be evaluated is scored according to the health level of the indicators: Assume the scores of each indicator are s1, s2, ..., s n The subjective and objective weights are w1, w2, ..., w n The overall health score H of the equipment to be evaluated is... e : H e =∑ i (w i s i ); In the formula, i∈[1,n], n represents the total number of indicators, w i Let s be the combined subjective and objective weights of the i-th indicator. i The score for the i-th indicator; (6c) Select equipment health assessment and present the scores of each indicator in the form of a radar chart: map each indicator name to one endpoint of the radar chart, and assign each indicator score to an endpoint s1, s2, ..., s2. n Mark the values ​​on the corresponding axes, and connect the marked points in sequence according to the index to form the distribution pattern of the radar chart. (6d) Calculate the total health score H e The concentric donut chart placed to the right of the radar chart includes four rings, H e Scores in [0,59], (60,79], (80,89], and [90,100] represent the range, pass, good, and excellent, respectively. The calculated H... e Hit the ring to obtain the evaluation result.

8. The comprehensive evaluation method for valve hall equipment status combining multispectral and data cube as described in claim 3, characterized in that: Step (2b) specifically includes the following steps in sequence: (2b1) Multi-scale image analysis is achieved through cascaded pyramid decomposition. First, the source image is decomposed into multiple scales using non-subsampled pyramid transformation. (2b2) ​​The high-frequency components are further decomposed in multiple directions using a non-subsampled directional filter bank to obtain sub-band images at different scales and in different directions; (2b3) The obtained sub-band image is processed by setting a threshold to filter out noise. A reasonable threshold is set to remove weak signals and noise in the image. (2b4) The image after non-subsampled contour wave transformation is reconstructed to obtain a denoised multispectral image, which improves the signal-to-noise ratio and clarity of the image and provides high-quality image data for subsequent feature extraction and analysis.

9. An electronic device, comprising: processor; as well as A memory storing computer program instructions that, when executed by the processor, cause the processor to perform the valve hall equipment condition comprehensive evaluation method combining multispectral and data cube as described in any one of claims 1-8.

10. A computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, cause the processor to perform the valve hall equipment condition comprehensive evaluation method combining multispectral and data cube as described in any one of claims 1-8.

Images