A power equipment temperature rise fault diagnosis method, system, device and medium

Abstract

The application discloses a power equipment temperature rise fault diagnosis method, system, device and medium, and comprises the following steps: acquiring temperature values of power equipment scattered points; reconstructing the temperature values of spatially dispersed sampling by a tensor block correlation matching method to obtain low-resolution three-dimensional temperature field reconstruction data; filling data of the low-resolution three-dimensional temperature field reconstruction data by a low-rank tensor super-resolution method to obtain high-resolution three-dimensional spatial temperature field data; and comparing the high-resolution three-dimensional spatial temperature field data with a differentiated threshold to perform power equipment temperature rise fault early warning according to a comparison result. The application improves the accuracy of power distribution switch cabinet temperature rise fault diagnosis.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power grid fault diagnosis, and particularly relates to a power equipment temperature rise fault diagnosis method, system, device and medium. BACKGROUND

[0002] The power distribution switch cabinet is responsible for distributing the power of a certain circuit of the upper-level power distribution equipment to the nearby load, and providing protection, monitoring and control for the load. The temperature online monitoring system thereof mainly comprises a temperature sensor, a temperature display instrument and system analysis software, and is one of important components of the power distribution switch cabinet. The temperature sensor is usually installed at the measured point in the cabinet and obtains the temperature value in a contact mode; and some infrared temperature sensors can be installed outside the cabinet and obtain the temperature value inside the power distribution switch cabinet in a non-contact mode.

[0003] At present, there are many defects in the research on the temperature rise monitoring and diagnosis of the power distribution switch cabinet. The space-time variation of the temperature field contains important flow information, but it is difficult to obtain a flow field with high time resolution and spatial resolution at the same time in actual measurement. For example, if only contact temperature sensors are used for measurement, a large number of sensors need to be installed to monitor the full amount of sensitive components, and in some working conditions, it is even impossible to install all the sensors. Sometimes, the installation of the sensors will disturb the temperature field, greatly affecting the accuracy of the measurement; if only non-contact temperature sensors are used for measurement, although the influence on the temperature field inside the power distribution switch cabinet can be reduced, it is difficult to accurately grasp the internal situation of the power distribution switch cabinet. SUMMARY

[0004] In order to realize non-invasive abnormal temperature rise state evaluation for the power distribution switch cabinet, the application provides a power equipment temperature rise fault diagnosis method, system, device and medium. The method can effectively support the abnormal temperature rise state evaluation requirement of the power distribution switch cabinet and improve the temperature rise fault diagnosis effect of the power distribution switch cabinet.

[0005] To achieve the above purpose, the application adopts the following technical solutions:

[0006] A power equipment temperature rise fault diagnosis method, comprising:

[0007] obtaining temperature values of dispersed points of the power equipment;

[0008] reconstructing the temperature values of the spatially dispersed sampling by a tensor block correlation matching method to obtain low-resolution three-dimensional temperature field reconstruction data;

[0009] filling data of the low-resolution three-dimensional temperature field reconstruction data by a low-rank tensor super-resolution method to obtain high-resolution three-dimensional spatial temperature field data;

[0010] The high-resolution three-dimensional temperature field data is compared with a differential threshold to perform power equipment temperature rise fault early warning based on the comparison result.

[0011] As a further improvement of the application, the temperature values are collected by a plurality of infrared temperature sensors arranged at intervals inside and outside the power equipment, the plurality of infrared temperature sensors collect temperature values of each position of the power equipment at the same time, the collected spatially dispersed temperature values are recorded in the form of a tensor, and the tensor is saved in the form of sparsification.

[0012] As a further improvement of the application, the temperature values collected by the tensor block association matching method are reconstructed to obtain low-resolution three-dimensional temperature field reconstruction data, including:

[0013] The temperature values to be reconstructed are represented in the form of a tensor I ∈ R s×t×l , the source area is C, the target area to be repaired is T, the label color of the target area T is a, the set of tensor blocks X(p) centered at point p and all points in the source area C is Ф C , the set of tensor blocks X(p) centered at point p and having a point falling in the target area T is Ф T , the boundary curve of the source area C and the target area T is δ T , the tensor block set centered at the pixel point on the boundary δ T , the tensor block set centered at the pixel point on the boundary δ T ,

[0014] According to the association of adjacent tensor blocks, the target area is repaired layer by layer, and the pixel value deviation of the overlapping area between the tensor blocks is constrained; the temperature value reconstruction is realized by repairing the target area T layer by layer from the outside to the inside;

[0015] In the repair process of the i-th layer, the tensor block X T with the highest priority in the tensor block set in the i-th layer is repaired first; secondly, each tensor block is repaired along the boundary until the pixel value deviation of the overlapping area between the tensor block and the repaired blocks in the neighborhood is less than the pre-set convergence threshold ε1, then the i-th layer algorithm converges, and the next layer repair begins.

[0016] As a further improvement of the application, the tensor block association matching method includes:

[0017] S1: determining the outermost tensor block set P T i of the target area T and the tensor block set Ф C of the source area C, and implementing the following steps until the i-th layer is repaired

[0018] S1.1: find the optimal repaired image block X in the set T ;

[0019] X T = argmax λ1||g(X(p))-aΛ‖0+(1-λ1)e(X(p))

[0020] s.t. X(p)∈P T t

[0021] where X T is the optimal repaired image block, λ1 is a constant between 0 and 1, g(X(p))-aΛ0 represents the number of pixels in the tensor block X(p) belonging to the source region, the g() function represents a function of mapping the tensor block to the entire space, X(p) is the tensor block within the source region C, a is a constant for identifying the target region, Λ is a full 1 matrix of the same size as g(X(p)), and e(X(p)) is a linear structure evaluation index of the tensor block X(p);

[0022] S1.2: for the to-be-repaired tensor block X T , find the most matched tensor block X C in the source region C using the following method to repair the missing region;

[0023]

[0024] S1.3: update the repaired region Ω←X C ∩T∪Ω and record the gray information g(X C ∩T) into the gray information g(Ω∪C) of the repaired tensor, update the source region C←C+Ω, update the target region T←T-Ω, and update μ1←cμ1;

[0025] if , go to S2; otherwise, select the next target tensor block as X T , and return to S1.2;

[0026] S2: update the tensor block layer number i←i+1, initialize the adjustment parameter μ←μ0, and return to S1;

[0027] Output: the gray information g(Ω∪C) of the repaired tensor, and a coarse repaired tensor I1 is obtained.

[0028] As a further improvement of the application, the low-rank tensor super-resolution method is used to fill data for low-resolution three-dimensional temperature field reconstruction data, and high-resolution three-dimensional spatial temperature field data is obtained, which comprises:

[0029] The low-resolution tensor of the 3D temperature field reconstruction data is extended into a high-resolution tensor, and the eight-neighbor linear representation coefficients β in the low-resolution tensor are used. k With the eight-neighbor linear representation coefficients in high-resolution tensors The low-rank matrix containing missing data is decomposed into a low-rank matrix A and a sparse matrix E. Repairing the low-rank matrix of the noisy matrix S involves:

[0030] Input a low-resolution tensor;

[0031] For a low-rank matrix of a noisy matrix S, fill the tensor intersection points with the known data of low-resolution tensor pixels and interpolation points.

[0032] For a low-rank matrix of a noisy matrix S, using low-resolution tensor pixels and intersection points as known data, perform secondary padding on the midpoints in the horizontal direction.

[0033] For a low-rank matrix of a noisy matrix S, using low-resolution tensor pixels and intersection points as known data, perform secondary padding on the middle points in the vertical direction.

[0034] Output the tensor of the entire repaired image.

[0035] As a further improvement of the present invention, the optimization model for repairing the low-rank matrix of the noisy matrix S is as follows:

[0036]

[0037] Where A is the repaired low-rank tensor, E is the noise tensor, ||·||1 represents the I1 norm of the target tensor, and ||·|| * Let S denote the nuclear norm of the matrix, λ² be the penalty factor, П denote the set of coordinates of known elements in S, and E ∏ =0 indicates that the values ​​of the noise tensor E at position П are all zero; Y is the Lagrange multiplier tensor, and μ2 is the contraction coefficient;

[0038] An optimized model for repairing the low-rank matrix of the noisy matrix S is solved to obtain high-resolution three-dimensional spatial temperature field data.

[0039] As a further improvement of the present invention, solving the optimization model for repairing the low-rank matrix of the noisy matrix S includes:

[0040] S11: Combining singular value decomposition with operator contraction, solve the following equation:

[0041] S11.1: Update

[0042] S11.2: update where Γ t (x) = sign(x) max{|x|-t, 0} represents a shrinkage operator;

[0043] S11.3: update

[0044] S11.4: if output directly; otherwise, jump to S12;

[0045] S12: update the unchanged position information E ∏ ←(E k+1 ) ∏ ;

[0046] S13: update the shrinkage coefficient μ 2(k+1) ←ρμ 2(k) ;

[0047] S14: update the Lagrange multiplier tensor Y k+1 ←Y k +μ 2(k) (S-A k+1 -E k+1 +E ∏ );

[0048] Output: repair the gray information of the tensor to obtain a fine repair tensor I2 = A;

[0049] According to the repaired tensor, a three-dimensional temperature field with high resolution is constructed.

[0050] As a further improvement of the application, the high-resolution three-dimensional spatial temperature field data is compared with a differentiated threshold to perform power equipment temperature rise fault early warning based on the comparison result, comprising:

[0051] Different temperature rise fault early warning thresholds are set for different components of the power equipment, and the temperature rise values are compared with the high-resolution three-dimensional spatial temperature field data, and when the local temperature rise exceeds the threshold, the temperature rise fault is diagnosed.

[0052] A power equipment temperature rise fault diagnosis system, comprising:

[0053] A numerical acquisition module for acquiring temperature values of power equipment dispersed points;

[0054] A first reconstruction module for reconstructing the temperature values of spatial dispersed sampling by a tensor block association matching method to obtain low-resolution three-dimensional temperature field reconstruction data;

[0055] A second reconstruction module is configured to perform data filling on the low-resolution three-dimensional temperature field reconstruction data by using a low-rank tensor super-resolution method, so as to obtain high-resolution three-dimensional spatial temperature field data.

[0056] A fault early warning module is configured to compare the high-resolution three-dimensional spatial temperature field data with a differential threshold, and perform power equipment temperature rise fault early warning based on the comparison result.

[0057] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the power equipment temperature rise fault diagnosis method when executing the computer program.

[0058] A computer readable storage medium stores a computer program, and the computer program implements the steps of the power equipment temperature rise fault diagnosis method when executed by a processor.

[0059] Compared with the prior art, the present application has the following beneficial effects:

[0060] The present application records the discrete temperature monitoring data of the power equipment as a tensor, obtains a lower resolution temperature field tensor by using a tensor block association matching method, and then fills in the default values of the low-resolution tensor based on a low-rank tensor super-resolution method, so that the low-resolution tensor becomes a high-resolution tensor, thereby realizing the reconstruction of a high-resolution three-dimensional spatial temperature field. According to the temperature rise early warning threshold of different components in space, the temperature rise fault diagnosis is performed, and the temperature rise fault diagnosis effect of the power distribution switch cabinet is improved. Specifically, the tensor block association matching method is proposed, the target area is repaired layer by layer according to the association of adjacent tensor blocks, the pixel value deviation of the overlapping area between the tensor blocks is constrained, the information change continuity of the adjacent area in the tensor is considered, and the data reconstruction accuracy is improved. The low-rank tensor super-resolution method is designed. The technology of obtaining a high-resolution tensor from a low-resolution tensor can increase the number of pixels according to the texture law of the tensor without changing the main contour information of the tensor, and ensures the gradual change characteristics of the tensor details. The power distribution switch cabinet temperature rise fault early warning mechanism based on the differential threshold is designed. Different temperature rise fault early warning thresholds are set for different components of the power distribution switch cabinet, such as wires, switches, windings, etc., the temperature rise values of the three-dimensional temperature field are compared, and the temperature rise fault diagnosis accuracy of the power distribution switch cabinet is improved. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 A flow chart of the power equipment temperature rise fault diagnosis method of the present application;

[0062] Figure 2 A schematic diagram of obtaining temperature values of scattered points outside the power distribution switch cabinet in step one;

[0063] Figure 3 A low-resolution temperature field reconstruction schematic diagram of a power distribution switch cabinet based on the tensor block correlation matching method of step two;

[0064] Figure 4 A low-rank matrix filling process schematic diagram;

[0065] Figure 5 A low-resolution temperature field reconstruction schematic diagram of a power distribution switch cabinet based on the low-rank tensor super-resolution method of step three;

[0066] Figure 6 A power distribution switch cabinet temperature rise fault early warning schematic diagram based on the spatial difference threshold of step four;

[0067] Figure 7 A three-dimensional reconstructed temperature field schematic diagram of a power distribution switch cabinet;

[0068] Figure 8 A comparison diagram of original data and compressed sensing processing results of the application; (a) original data; (b) compressed sensing processing results; (c) heat source condition;

[0069] Figure 9 A power equipment temperature rise fault diagnosis system of the application;

[0070] Figure 10 An electronic device schematic diagram of the application. DETAILED DESCRIPTION

[0071] In order for those skilled in the art to better understand the application scheme, the technical solutions in the embodiments of the application will be described clearly and completely below in conjunction with the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the application.

[0072] It should be noted that the terms "first", "second", etc. in the specification and claims of the application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0073] The application belongs to the field of temperature field analysis and temperature rise fault diagnosis of power equipment, and particularly relates to a temperature field reconstruction and temperature rise fault diagnosis method based on compressed sensing of a power distribution switch cabinet.

[0074] The application provides a power equipment temperature rise fault diagnosis method, which comprises the following steps of:

[0075] obtaining temperature values of dispersed points of the power equipment;

[0076] reconstructing the temperature values of the spatial dispersed sampling by a tensor block correlation matching method to obtain low-resolution three-dimensional temperature field reconstruction data;

[0077] filling data of the low-resolution three-dimensional temperature field reconstruction data by a low-rank tensor super-resolution method to obtain high-resolution three-dimensional spatial temperature field data;

[0078] comparing the high-resolution three-dimensional spatial temperature field data with a differentiated threshold to perform power equipment temperature rise fault early warning according to a comparison result.

[0079] The method reconstructs the temperature field of the power distribution switch cabinet based on compressed sensing, first obtains dispersed temperature sensing parameters outside the power distribution switch cabinet, then obtains the temperature field by combining a tensor block correlation matching method, and finally performs high-resolution processing on the low-resolution tensor based on a low-rank tensor super-resolution method, so as to realize high-resolution temperature field reconstruction of the three-dimensional space. According to the temperature rise early warning threshold of different components in the space, temperature rise fault diagnosis is performed.

[0080] The method is based on the dispersed temperature sensing parameters outside the power distribution switch cabinet, combines the tensor block correlation matching and the low-rank tensor super-resolution method, finds suitable units in the temperature field tensor dictionary from outside to inside to fill the target area, and realizes internal temperature field reconstruction of the power distribution switch cabinet. The method has good fitting effect on the two-dimensional temperature profile and the three-dimensional temperature field, and can effectively support the abnormal temperature rise state evaluation requirement of the power distribution switch cabinet.

[0081] The power distribution switch cabinet temperature rise fault diagnosis method based on compressed sensing of the application can improve the application technical level and application effect of temperature analysis and temperature rise fault diagnosis of the power distribution switch cabinet from three aspects.

[0082] First, the temperature values of the spatial dispersed sampling are reconstructed by the tensor block correlation matching method to obtain low-resolution spatial data, and the problem of reconstructing the three-dimensional spatial temperature field according to a small amount of dispersed sampling temperature data is solved.

[0083] Second, the low-resolution three-dimensional temperature field reconstruction data are filled by the low-rank tensor super-resolution method to obtain high-resolution three-dimensional spatial temperature field data, and the problem of constructing the high-resolution three-dimensional spatial temperature field by relying on the dispersed sampling temperature data is solved.

[0084] Thirdly, by comparing the temperature of different regions in space with the differentiated threshold, the fine evaluation of the temperature rise fault of the power distribution switch cabinet is realized.

[0085] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0086] Embodiments

[0087] Taking the three-dimensional temperature field reconstruction and temperature rise fault diagnosis of the power distribution switch cabinet in the power field as an example, the present application realizes the above-mentioned purposes through the following technical solutions:

[0088] Step one: obtaining the temperature values of the scattered points outside the power distribution switch cabinet

[0089] Figure 2 To obtain the temperature values of the scattered points outside the power distribution switch cabinet

[0090] A plurality of infrared temperature sensors are scattered and arranged at intervals of 10cm to 30cm inside and outside the power distribution switch cabinet, and the temperature values of each position of the power distribution switch cabinet at the same time are collected in a non-contact manner, and the temperature values are saved in the form of sparse tensor. The collected space discrete temperature values are recorded in the form of tensor.

[0091] Step two: low-resolution temperature field reconstruction of the power distribution switch cabinet based on the tensor block association matching method

[0092] Figure 3 Flow chart for low-resolution temperature field reconstruction of the power distribution switch cabinet based on the tensor block association matching method; the temperature discrete test data to be reconstructed is expressed in the form of tensor I ∈ R s×t×l , let the source region be C, the target region to be repaired be T, and the label color of the region T be a. Let the set of tensor blocks X(p) centered at point p and all points in the source region C be Ф C The set of tensor blocks X(p) centered at point p and having points falling in the target region T is Ф T The boundary curve of the source region C and the target region T is δ T The tensor block set centered at the pixel point on the boundary δ T P T ,

[0093] According to the correlation of adjacent tensor blocks, the target region is repaired layer by layer, and the pixel value deviation of the overlapping region between the tensor blocks is constrained. Through the repair of the target region T from the outside to the inside layer by layer, the data reconstruction is realized. In the repair process of the ith layer, first, the highest priority to-be-repaired block X T is repaired in the ith layer tensor block set; second, each tensor block is repaired along the boundary in turn until the pixel value deviation of the overlapping region between it and the repaired block in the neighborhood is less than the preset convergence threshold ε1, then the ith layer algorithm converges, and the next layer repair begins. The specific process of the tensor block correlation matching repair method is as follows:

[0094] Step 1: Determine the outermost tensor block set of the target region T and the tensor block set Ф in the source region C C , and implement the following steps until the ith layer repair is completed

[0095] Step 1.1: Find the optimal repair image block X in the set T according to formula (1)

[0096] X T = argmax λ1||g(X(p))-aΛ‖0+(1-λ1)e(X(p))

[0097] s.t.X(p)∈P T t

[0098] In the formula, X T is the optimal repair image block, λ1 is a constant between 0 and 1, ||g(X(p))-aΛ||0 represents the number of pixels in the tensor block X(p) belonging to the source region, g() function represents a function of mapping the tensor block to the entire space, X(p) is a tensor block in the source region C, a is a constant for identifying the target region, Λ is a full 1 matrix with the same size as g(X(p)), and e(X(p)) is a linear structure evaluation index of the tensor block X(p);

[0099] Step 1.2: For the to-be-repaired tensor block X T , find the most matched tensor block X C in the source region C using formula (2) to repair the missing region

[0100]

[0101] Step 1.3: Update the repaired region Ω←X C ∩T∪Ω and record the gray scale information g(X CUpdate the source region C←C+Ω, the target region T←T-Ω, and μ1←cμ1. If go to Step 2; otherwise, select the next target tensor block in the clockwise (counterclockwise) direction as X T , and return to Step 1.2;

[0102] Step 2: Update the tensor block layer i←i+1, initialize the adjustment parameter μ←μ0, and return to Step 1;

[0103] Output: the gray information g(Ω∪C) of the repaired tensor, and obtain the rough repaired tensor I1.

[0104] Step 3: Reconstruction of the low-resolution temperature field of the power distribution switch cabinet based on the low-rank tensor super-resolution method

[0105] Figure 5 Reconstruction of the low-resolution temperature field of the power distribution switch cabinet based on the low-rank tensor super-resolution method

[0106] Step 3: Extend the rough repaired low-resolution tensor obtained in Step 2 into a high-resolution tensor (as shown in Figure 4 ). Figure 4 Flowchart for filling a low-rank matrix; in order to obtain a high-resolution tensor from a low-resolution tensor, this method increases the number of pixels according to the texture characteristics of the tensor without changing the main profile information of the tensor, so as to ensure the gradual change characteristics of the details of the tensor. Taking the three-dimensional tensor in the following figure as an example, a 4*4*4 low-resolution tensor is generated into a 7*7*7 high-resolution tensor, and the black points represent known pixels and the gray points represent unknown pixels.

[0107] According to the similarity between the low-resolution and high-resolution tensors, the linear representation coefficient β k of the eight-neighborhood in the low-resolution tensor should be basically equal to the linear representation coefficient β of the eight-neighborhood in the high-resolution tensor. A low-rank matrix with missing data can be decomposed into the sum of a low-rank matrix A and a sparse matrix E (which can be regarded as a noise matrix).

[0108] As shown in Figure 5 , the repairing method includes the following steps:

[0109] Input the low-resolution tensor;

[0110] Fill the low-rank matrix with the noisy matrix S, using the low-resolution tensor pixels and the interpolation points as known data, and filling the cross-position points of the tensor;

[0111] Fill the low-rank matrix with the noisy matrix S, using the low-resolution tensor pixels and the cross-position points as known data, and filling the middle points in the horizontal direction twice.

[0112] For a low-rank matrix of a noisy matrix S, using low-resolution tensor pixels and intersection points as known data, perform secondary padding on the middle points in the vertical direction.

[0113] Output the tensor of the entire repaired image.

[0114] The optimization problem of repairing missing data in the noisy matrix S can be described as follows:

[0115] A = arg minλ2||E||1 +||A|| *

[0116] stS = A + E, E ∏ =0,

[0117] Where A is the repaired low-rank tensor, E is the noise tensor, ||·||1 represents the I1 norm of the target tensor, and ||·|| * Let S denote the nuclear norm of the matrix, λ² be the penalty factor, П denote the set of coordinates of known elements in S, and E ∏ =0 indicates that the values ​​of the noise tensor E at position П are all zero. Applying the augmented Lagrange multiplier method to solve the optimization problem given in equation (3), the repaired tensor A is obtained. Rewriting the original problem as an augmented Lagrange multiplication equation, we get:

[0118]

[0119] Where Y is the Lagrange multiplier tensor and μ2 is the shrinkage coefficient, the detailed repair process for noisy low-rank tensor filling is as follows:

[0120] Step 1: By combining Singular Value Decomposition (SVD) with the operator contraction method, the following equations can be solved:

[0121] Step 1.1: Update

[0122] Step 1.2: Update Where Γ t (x) = sign(x)·max{|x|-t, 0} represents the contraction operator;

[0123] Step 1.3: Update

[0124] Step 1.4: If Output directly; otherwise, jump to step 2.

[0125] Step 2: Update the unchanged location information E ∏ ←(Ek+1 ∏ ;

[0126] Step 3: update shrinkage coefficient mu 2(k+1) ← p mu 2(k) ;

[0127] Step 4: update Lagrange multiplier tensor Y k+1 ← Y k + mu 2(k) (S-A k+1 -E k+1 +E ∏ );

[0128] Output: repair the gray information of the tensor to obtain a fine repair tensor I2=A.

[0129] According to the repaired tensor, a three-dimensional temperature field with high resolution can be constructed.

[0130] Step four: power distribution switch cabinet temperature rise fault early warning based on differentiated threshold

[0131] Figure 6 It is a flow chart of power distribution switch cabinet temperature rise fault early warning based on spatial differentiated threshold. Different temperature rise fault early warning thresholds are set for different components of the power distribution switch cabinet, such as wires, switches, windings, etc. The temperature rise values of the three-dimensional temperature field are compared. When the local temperature rise exceeds the threshold, it is diagnosed as a temperature rise fault.

[0132] Figure 7 It is a schematic diagram of the three-dimensional reconstructed temperature field of the power distribution switch cabinet. Through the processing procedures of steps one to four, based on the temperature rise data of the discrete monitoring of the power distribution switch cabinet, the three-dimensional temperature field of the power distribution switch cabinet can be obtained, and the temperature rise fault of the power distribution switch cabinet can be diagnosed.

[0133] Based on the above embodiment description, the discrete temperature monitoring data of the power distribution switch cabinet can be recorded as a tensor, a lower resolution temperature field tensor can be obtained by the tensor block association matching method, and the default value calculation filling of the low resolution tensor can be performed based on the low rank tensor super resolution method, so that the low resolution tensor becomes high resolution, thereby realizing the reconstruction of the three-dimensional space high resolution temperature field. According to the temperature rise early warning threshold of different components in space, the temperature rise fault diagnosis is carried out, and the effect of the power distribution switch cabinet temperature rise fault diagnosis is improved.

[0134] The advantages of the method of the present application are as follows:

[0135] Firstly, the tensor block association matching method is proposed. According to the association of adjacent tensor blocks, the target area is repaired layer by layer, and the pixel value deviation of the overlapping area between the tensor blocks is constrained. The continuity of the information change of the adjacent area in the tensor is considered, and the data reconstruction accuracy is improved.​

[0136] Secondly, a low-rank tensor super-resolution method is designed. The technology of obtaining a high-resolution tensor from a low-resolution tensor can increase the number of pixels according to the texture law of the tensor, without changing the main contour information of the tensor, and ensure the gradual change characteristics of the details of the tensor.

[0137] Thirdly, a power distribution switch cabinet temperature rise fault early warning mechanism based on a differentiated threshold is designed. Different temperature rise fault early warning thresholds are set for different components of the power distribution switch cabinet, such as wires, switches, windings, etc., the temperature rise numerical comparison of the three-dimensional temperature field is performed, and the accuracy of the power distribution switch cabinet temperature rise fault diagnosis is improved.

[0138] Figure 8 The figure (a) is the original data, (b) is the processing result of the compressed sensing, and (c) is the heat source condition. It can be seen from the comparison that the method improves the accuracy of the power distribution switch cabinet temperature rise fault diagnosis.

[0139] As shown in Figure 9 The application further provides a power equipment temperature rise fault diagnosis system, which comprises:

[0140] A numerical acquisition module is configured to acquire temperature numerical values of dispersed points of the power equipment.

[0141] A first reconstruction module is configured to reconstruct the temperature numerical values of the spatial dispersed sampling by using a tensor block correlation matching method, to obtain low-resolution three-dimensional temperature field reconstruction data.

[0142] A second reconstruction module is configured to fill data of the low-resolution three-dimensional temperature field reconstruction data by using a low-rank tensor super-resolution method, to obtain high-resolution three-dimensional spatial temperature field data.

[0143] A fault early warning module is configured to compare the high-resolution three-dimensional spatial temperature field data with the differentiated threshold, to perform power equipment temperature rise fault early warning according to the comparison result.

[0144] The application effectively saves the temperature rise data in the form of a tensor, proposes a tensor block correlation matching method, repairs the target area layer by layer according to the correlation of adjacent tensor blocks, and constrains the pixel value deviation of the overlapping area between the tensor blocks, thereby improving the data reconstruction accuracy. The low-rank tensor super-resolution method is designed to obtain a high-resolution tensor from a low-resolution tensor, increase the number of pixels according to the texture law of the tensor, and ensure the gradual change characteristics of the details of the tensor. The power distribution switch cabinet temperature rise fault early warning mechanism based on the differentiated threshold is designed, the differentiated temperature rise threshold is compared for different components of the power distribution switch cabinet, and the accuracy of the power distribution switch cabinet temperature rise fault diagnosis is improved.

[0145] In the numerical acquisition module, the temperature values ​​are collected by multiple infrared temperature sensors that are distributed at intervals inside and outside the power equipment. The multiple infrared temperature sensors collect the temperature values ​​at various locations of the power equipment at the same time. The collected spatially discrete temperature values ​​are recorded in the form of tensors and stored in the form of sparse tensors.

[0146] In the first reconstruction module, the process of reconstructing the spatially discrete sampled temperature values ​​using a tensor block association matching method to obtain low-resolution three-dimensional temperature field reconstruction data includes:

[0147] The temperature value to be reconstructed is represented as a tensor I∈R. s×t×l Let the source region be C, the target region to be repaired be T, and the color of region T be a; let the set of tensor blocks X(p) centered at point p and containing all points within the source region C be Ф. C Let Ф be the set of tensor blocks X(p) centered at point p that contain points within the target region T. T The boundary curve between the source region C and the target region T is μ. T , with boundary δ T The set of tensor blocks centered on the pixels on the array is P. T ,

[0148] The target region is repaired layer by layer based on the correlation between adjacent tensor blocks, and the pixel value deviation in the overlapping area between tensor blocks is constrained; the temperature value is reconstructed by repairing the target region T layer by layer from the outside to the inside.

[0149] During the repair process of the i-th layer, the highest priority block X in the tensor block set of the i-th layer is repaired first. T Secondly, each tensor block is repaired sequentially along the boundary until the pixel value deviation between it and the neighboring repaired block is less than the preset convergence threshold ε1. Then the i-th layer algorithm converges and the next layer repair begins.

[0150] In the second reconstruction module, the step of using a low-rank tensor super-resolution method to fill in the low-resolution three-dimensional temperature field reconstruction data to obtain high-resolution three-dimensional spatial temperature field data includes:

[0151] The low-resolution tensor of the 3D temperature field reconstruction data is extended into a high-resolution tensor, and the eight-neighbor linear representation coefficients β in the low-resolution tensor are used. k With the eight-neighbor linear representation coefficients in high-resolution tensors The low-rank matrix containing missing data is decomposed into a low-rank matrix A and a sparse matrix E. Repairing the low-rank matrix of the noisy matrix S involves:

[0152] Input low resolution tensor;

[0153] The low rank matrix of the noisy matrix S is filled with the low resolution tensor pixel points and the interpolation points as known data.

[0154] The low rank matrix of the noisy matrix S is filled with the low resolution tensor pixel points and the interpolation points as known data.

[0155] The low rank matrix of the noisy matrix S is filled with the low resolution tensor pixel points and the interpolation points as known data.

[0156] Output the repaired tensor.

[0157] In the fault early warning module, the high-resolution three-dimensional space temperature field data is compared with the differentiated threshold value to perform power equipment temperature rise fault early warning based on the comparison result, and the method comprises the following steps:

[0158] Different temperature rise fault early warning thresholds are set for different components of the power equipment, and the temperature rise values are compared with the high-resolution three-dimensional space temperature field data, and when the local temperature rise exceeds the threshold value, the temperature rise fault is diagnosed.

[0159] As shown in Figure 10 The present application provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the power equipment temperature rise fault diagnosis method when executing the computer program.

[0160] The power equipment temperature rise fault diagnosis method comprises the following steps:

[0161] Obtain the temperature values of the dispersed points of the power equipment.

[0162] Reconstruct the temperature values of the spatial dispersed sampling by a tensor block correlation matching method to obtain low-resolution three-dimensional temperature field reconstruction data.

[0163] Fill the data of the low-resolution three-dimensional temperature field reconstruction data by a low-rank tensor super-resolution method to obtain high-resolution three-dimensional space temperature field data.

[0164] Compare the high-resolution three-dimensional space temperature field data with the differentiated threshold value to perform power equipment temperature rise fault early warning based on the comparison result.

[0165] The present application also provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the power equipment temperature rise fault diagnosis method when executed by a processor.

[0166] The power equipment temperature rise fault diagnosis method comprises the following steps:

[0167] Obtaining temperature values of power equipment dispersed points;

[0168] Reconstructing the temperature values of the spatial dispersed sampling by a tensor block correlation matching method to obtain low-resolution three-dimensional temperature field reconstruction data;

[0169] Filling data of the low-resolution three-dimensional temperature field reconstruction data by a low-rank tensor super-resolution method to obtain high-resolution three-dimensional spatial temperature field data;

[0170] Comparing the high-resolution three-dimensional spatial temperature field data with a differentiated threshold to perform power equipment temperature rise fault early warning according to a comparison result.

[0171] Those skilled in the art will understand that embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0172] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0173] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0174] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0175] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the technical solutions of the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A method for diagnosing temperature rise faults in power equipment, characterized in that, include: Obtain temperature values ​​at distributed locations of power equipment; The temperature values ​​sampled in spatial discrete form are reconstructed using the tensor block association matching method to obtain low-resolution three-dimensional temperature field reconstruction data. By using the low-rank tensor super-resolution method, data filling is performed on the low-resolution three-dimensional temperature field reconstruction data to obtain high-resolution three-dimensional spatial temperature field data. High-resolution three-dimensional spatial temperature field data is compared with differentiated thresholds to provide early warning of temperature rise faults in power equipment. The method of reconstructing the spatially discrete sampled temperature values ​​using tensor block association matching to obtain low-resolution three-dimensional temperature field reconstruction data includes: The temperature value to be reconstructed is represented as a tensor. Let the source region be The target area to be repaired is ,area The marker color is ; Order with points Centered on the source region and all points are within the source region Tensor blocks inside The set is , with point Centered on a point that falls within the target area Tensor blocks inside The set is Source region With the target area The boundary curve is , with the boundary The set of tensor blocks centered on the pixels above is , ; The target region is repaired layer by layer based on the correlation between adjacent tensor blocks, and the pixel value deviation in the overlapping areas between tensor blocks is constrained; by targeting the region... T Temperature values ​​are reconstructed by repairing layer by layer from the outside in. In the case of the During the repair process of the layer, the first layer is repaired. The highest priority block to be repaired in the set of tensor blocks within the layer. Secondly, each tensor block is repaired sequentially along the boundary until the pixel value deviation between it and the overlapping area of ​​the neighboring repaired blocks is less than a preset convergence threshold. Then for the first Once the layer algorithm converges, begin the next layer of repair. The tensor block association matching method includes: S1: Determine the outermost set of tensor blocks for the target region T. and source region Tensor block set inside Follow the steps below until the [number]th [number]. Layer repair completed : S1.1: In the set according to the following method Find the image patch with the highest priority for repair. ; In the formula, The image patch with the highest priority for repair, where λ1 is a constant between 0 and 1. Tensor blocks X The number of pixels belonging to the source region in (p) g The () function represents a function that maps a tensor block to the entire space. X (p) represents the source region. Tensor blocks within, a A constant used to identify the target region. Λ To and g ( X ( p A matrix of all ones of the same size. For tensor blocks X ( p linear structure evaluation index; S1.2: For the tensor block to be repaired The following method is used in the source region Find the best matching tensor block Repair the damaged area; S1.3: Update fixed areas And record grayscale information To repair the grayscale information of the tensor In the middle, update the source region Update target area ,renew ; like If the target tensor block is selected, proceed to S2; otherwise, the next target tensor block is selected as... Return to S1.2; S2: Update the tensor block level Initialize adjustment parameters Return to S1; Output: Repair the grayscale information of the tensor The coarse repair tensor is obtained. .

2. The method for diagnosing temperature rise faults in power equipment according to claim 1, characterized in that, The temperature values ​​are collected by multiple infrared temperature sensors that are distributed at intervals inside and outside the power equipment. The multiple infrared temperature sensors collect the temperature values ​​at various locations of the power equipment at the same time. The collected spatially discrete temperature values ​​are recorded in the form of tensors and stored in the form of sparse tensors.

3. The method for diagnosing temperature rise faults in power equipment according to claim 1, characterized in that, The method of using low-rank tensor super-resolution to fill in low-resolution three-dimensional temperature field reconstruction data to obtain high-resolution three-dimensional spatial temperature field data includes: The low-resolution tensor of the 3D temperature field reconstruction data is extended into a high-resolution tensor, and the coefficients in the low-resolution tensor are represented by an eight-neighbor linear representation. With the eight-neighbor linear representation coefficients in high-resolution tensors A low-rank matrix with missing data can be decomposed into a low-rank matrix. A and a sparse matrix E The sum of the sums for the noisy matrix S Repairing low-rank matrices includes: Input a low-resolution tensor; For noisy matrices S The low-rank matrix is ​​filled with the tensor intersection points using the low-resolution tensor pixels and interpolation points as known data. For noisy matrices S The low-rank matrix is ​​filled twice with the middle points in the horizontal direction, using the low-resolution tensor pixels and intersection points as known data. For noisy matrices S The low-rank matrix is ​​filled twice with the middle points in the vertical direction, using the low-resolution tensor pixels and intersection points as known data. Output the tensor of the entire repaired image.

4. The method for diagnosing temperature rise faults in power equipment according to claim 3, characterized in that, The pair of noisy matrices S The optimization model for repairing low-rank matrices is as follows: in, For the repaired low-rank tensor, For noise tensor, Representing the target tensor Norm, The nuclear norm of a matrix is ​​represented by its nucleus. П represents the penalty factor. S The set of coordinates of known elements in the set. Represents the noise tensor The values ​​at position П are all zero; Y For Lagrange multiplier tensors, It is the coefficient of shrinkage; For noisy matrices S The optimized model, which repairs the low-rank matrix, is solved to obtain high-resolution three-dimensional spatial temperature field data.

5. The method for diagnosing temperature rise faults in power equipment according to claim 4, characterized in that, The pair of noisy matrices S The optimization model for repairing low-rank matrices is solved, including: S11: Combining singular value decomposition with operator contraction, solve the following equation: ; S11.1: Update ; S11.2: Update , in This represents the contraction operator; S11.3: Update ; S11.4: If Output directly; otherwise, jump to S12. S12: Update unchanged location information ; S13: Update the shrinkage coefficient ; S14: Update the Lagrange multiplier tensor ; Output: The grayscale information of the tensor is repaired to obtain a finely repaired tensor. ; Based on the tensor obtained from the repair, a higher resolution three-dimensional temperature field is constructed.

6. The method for diagnosing temperature rise faults in power equipment according to claim 1, characterized in that, The comparison between high-resolution three-dimensional spatial temperature field data and differential thresholds, and the resulting comparison for early warning of temperature rise faults in power equipment, includes: Different temperature rise fault warning thresholds are set for different components of power equipment, and the temperature rise values ​​are compared with high-resolution three-dimensional spatial temperature field data. When the local temperature rise exceeds the threshold, it is diagnosed as a temperature rise fault.

7. A power equipment temperature rise fault diagnosis system, based on the power equipment temperature rise fault diagnosis method according to any one of claims 1 to 6, characterized in that, include: The data acquisition module is used to acquire temperature values ​​at various locations of power equipment. The first reconstruction module is used to reconstruct the spatially discrete sampled temperature values ​​using a tensor block association matching method to obtain low-resolution three-dimensional temperature field reconstruction data. The second reconstruction module is used to fill in the low-resolution three-dimensional temperature field reconstruction data with data using the low-rank tensor super-resolution method, so as to obtain high-resolution three-dimensional spatial temperature field data. The fault early warning module is used to compare high-resolution three-dimensional spatial temperature field data with differentiated thresholds to provide early warning of temperature rise faults in power equipment based on the comparison results.

8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the power equipment temperature rise fault diagnosis method according to any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the electrical equipment temperature rise fault diagnosis method according to any one of claims 1 to 6.

Images