CN112419243A - Power distribution room equipment fault identification method based on infrared image analysis - Google Patents
Power distribution room equipment fault identification method based on infrared image analysis Download PDFInfo
- Publication number
- CN112419243A CN112419243A CN202011250191.9A CN202011250191A CN112419243A CN 112419243 A CN112419243 A CN 112419243A CN 202011250191 A CN202011250191 A CN 202011250191A CN 112419243 A CN112419243 A CN 112419243A
- Authority
- CN
- China
- Prior art keywords
- infrared image
- distribution room
- image
- room equipment
- method based
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000009826 distribution Methods 0.000 title claims abstract description 53
- 238000010191 image analysis Methods 0.000 title claims abstract description 33
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000003745 diagnosis Methods 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims description 13
- 238000010586 diagram Methods 0.000 claims description 12
- 239000013598 vector Substances 0.000 claims description 10
- 230000009467 reduction Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 239000010749 BS 2869 Class C1 Substances 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000005065 mining Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 claims 1
- 239000000203 mixture Substances 0.000 claims 1
- 239000002994 raw material Substances 0.000 claims 1
- 230000011218 segmentation Effects 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 6
- 238000003708 edge detection Methods 0.000 description 6
- 238000000605 extraction Methods 0.000 description 6
- 230000004913 activation Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000011176 pooling Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000003709 image segmentation Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000004020 conductor Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000012212 insulator Substances 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000003921 oil Substances 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013530 stochastic neural network Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000004804 winding Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Primary Health Care (AREA)
- Operations Research (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Entrepreneurship & Innovation (AREA)
- Water Supply & Treatment (AREA)
- Molecular Biology (AREA)
- Multimedia (AREA)
- Public Health (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
- Radiation Pyrometers (AREA)
Abstract
The invention discloses a power distribution room equipment fault identification method based on infrared image analysis, which comprises the steps of collecting an infrared image of power distribution room equipment and carrying out image preprocessing on the infrared image; identifying power distribution room equipment in the infrared image based on a convolutional neural network; performing image registration on the infrared image by combining the equipment; and constructing a depth confidence network model, and carrying out fault diagnosis on the registration image. The invention can accurately identify each power distribution room device and identify the fault type of the device.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a power distribution room equipment fault recognition method based on infrared image analysis.
Background
The power distribution room is an important component of the power system and plays an important role. With the development and construction of smart grids, state data, image monitoring data and environmental meteorological data of power systems are gradually integrated and shared on a unified platform.
When fault diagnosis and analysis are carried out on the power distribution room equipment, the constructed model is required to be capable of displaying potential fault information existing in the equipment in time. Generally, potential faults of the transformer can be divided into internal faults and external faults, and the external faults of the cabinet mainly comprise external connection faults of conductors, faults of a cooling device and an oil way system and magnetic leakage faults; internal failures of cabinets are mainly caused by deficiencies and defects of internal equipment such as coils, cores and leads.
Current methods are not able to detect faults inside the cabinet directly, but different internal fault defects will result in different distributions of thermal conditions across the surface of the cabinet, so by analyzing the thermal condition profile displayed outside the cabinet, it is also possible to roughly analyze the type of fault present in the equipment.
The power distribution room equipment operation state analysis based on image analysis has been studied, for example, a convolutional neural network is adopted to extract the characteristics of an infrared image, and then the infrared image is aggregated by using the vector of a local aggregation descriptor; detecting faults occurring in the solar photovoltaic system by adopting a BP neural network; however, the current research mainly considers the decomposition of the device image, or only uses the RGB image data to classify the working state of the device, and there is no specific implementation plan for the complete closed-loop analysis system.
Disclosure of Invention
This section is for the purpose of summarizing some aspects of embodiments of the invention and to briefly introduce some preferred embodiments. In this section, as well as in the abstract and the title of the invention of this application, simplifications or omissions may be made to avoid obscuring the purpose of the section, the abstract and the title, and such simplifications or omissions are not intended to limit the scope of the invention.
The present invention has been made in view of the above-mentioned conventional problems.
Therefore, the invention provides a power distribution room equipment fault identification method based on infrared image analysis, which solves the problems of no image identification in a complete closed-loop analysis system and error and uncertainty caused by manually reading an infrared image by combining a convolutional neural network and a deep confidence network.
In order to solve the technical problems, the invention provides the following technical scheme: acquiring an infrared image of power distribution room equipment, and performing image preprocessing on the infrared image; identifying power distribution room equipment in the infrared image based on a convolutional neural network; performing image registration on the infrared image by combining the equipment; and constructing a depth confidence network model, and carrying out fault diagnosis on the registration image.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the image preprocessing comprises the step of carrying out noise reduction processing on the infrared image by using a mean value filtering method; splicing the infrared images subjected to the noise reduction treatment through characteristic matching; and dividing the spliced infrared image by using an OSTU algorithm.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the splicing comprises the steps of defining a characteristic point F (x, y) of the infrared image after the noise reduction treatment, enabling the main direction of the characteristic point F to be a coordinate axis, and rotating the coordinate of the characteristic point:
taking the feature point as a center, and obtaining a feature vector by calculating the gradient value and the direction of the region where the feature point is located; and performing the feature matching by using the feature vector.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the gradient values may include, for example,
wherein L (x, y) is the ratio of the feature points.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the directions include that the direction of the light beam comprises,
as a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: defining X as an L-level gray image, and dividing all pixels in the image into a target class C0 and a non-target class C1 according to a threshold k, wherein the gray value range of all pixels in the C0 domain is [0, k-1], and the gray value range of all pixels in the C1 domain is [ k, L-1 ]; the proportion of the total area occupied by the pixels of C0 and C1 is respectively as follows:
ω1＝1-ω0
wherein, PiThe area proportion of the ith pixel point in the C0 domain is that i is 0, 1, 2, …, m; defining the maximum variance between the C0 and C1 classes as:
δ2(k)＝ω0(μ-μ0)2+ω1(μ-μ1)2
wherein μ is the average gray scale, μ0Is the variance of C0, μ1Is the variance of C1.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the convolutional neural network comprises convolutional layers
The calculation formula of (a) is as follows:
wherein M isjFor the size of the input image, i is the step size, j is the height of the image, l is the width of the image,
is the number of the convolution kernels and is,
is the size of the convolution kernel and,
is the number of channels.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the image registration comprises that A is assumed to be an image with a correct position of the equipment, and B is assumed to be an image to be registered; taking any point Q (m, n) on the B diagram and any point R (u, v) on the A diagram, and respectively substituting the points into a registration equation:
and obtaining the values of a and B by solving the registration equation, traversing all pixel points of the B image and substituting the pixel points into the registration equation, and further obtaining all registration pixel points.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the deep belief network model comprises, pre-training: pre-training the Boltzmann machine layer by layer under the unsupervised condition, and deeply mining hidden characteristic information of data; fine adjustment: training and stacking the next layer of the Boltzmann machine, combining the labels with the samples for use, and performing supervised adjustment by adopting back propagation to realize fault classification.
As a preferred embodiment of the power distribution room equipment fault identification method based on infrared image analysis of the present invention, wherein: the back propagation comprises the step of updating the network parameters of the deep belief network model by adopting the back propagation strategy, and a cost function is defined as follows:
wherein E is the average square error, N is the number of hidden elements,
and XiRepresenting the output of the output layer and the ideal output respectively,i is the sample index, (W)l,bl) Representing the weights to be learned and the bias parameters at level l.
The invention has the beneficial effects that: by establishing the feature vectors at the feature points, each power distribution room device can be identified more accurately; the intelligent classification of different power distribution room equipment faults is realized by establishing a deep confidence network model, and the identification precision is high.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise. Wherein:
fig. 1 is a schematic flowchart of a method for identifying a fault of a power distribution room device based on infrared image analysis according to a first embodiment of the present invention;
fig. 2 is a schematic diagram illustrating a method for identifying faults of power distribution room equipment based on infrared image analysis according to a maximum pooling method of the first embodiment of the present invention;
fig. 3 is a schematic structural diagram of a restricted boltzmann machine of a power distribution room equipment fault identification method based on infrared image analysis according to a first embodiment of the present invention;
fig. 4 is a schematic structural diagram of a deep confidence network of a power distribution room equipment fault identification method based on infrared image analysis according to a first embodiment of the present invention;
fig. 5 is a schematic diagram illustrating a classification result of power distribution room equipment images according to a power distribution room equipment fault identification method based on infrared image analysis according to a second embodiment of the present invention;
fig. 6 is a schematic diagram illustrating a test result of a bushing under a normal condition in a method for identifying a fault of a distribution room equipment based on infrared image analysis according to a second embodiment of the present invention;
fig. 7 is a schematic diagram illustrating a test result of abnormal casing state in a distribution room equipment fault identification method based on infrared image analysis according to a second embodiment of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, specific embodiments accompanied with figures are described in detail below, and it is apparent that the described embodiments are a part of the embodiments of the present invention, not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making creative efforts based on the embodiments of the present invention, shall fall within the protection scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways than those specifically described and will be readily apparent to those of ordinary skill in the art without departing from the spirit of the present invention, and therefore the present invention is not limited to the specific embodiments disclosed below.
Furthermore, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
The present invention will be described in detail with reference to the drawings, wherein the cross-sectional views illustrating the structure of the device are not enlarged partially in general scale for convenience of illustration, and the drawings are only exemplary and should not be construed as limiting the scope of the present invention. In addition, the three-dimensional dimensions of length, width and depth should be included in the actual fabrication.
Meanwhile, in the description of the present invention, it should be noted that the terms "upper, lower, inner and outer" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of describing the present invention and simplifying the description, but do not indicate or imply that the referred device or element must have a specific orientation, be constructed in a specific orientation and operate, and thus, cannot be construed as limiting the present invention. Furthermore, the terms first, second, or third are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
The terms "mounted, connected and connected" in the present invention are to be understood broadly, unless otherwise explicitly specified or limited, for example: can be fixedly connected, detachably connected or integrally connected; they may be mechanically, electrically, or directly connected, or indirectly connected through intervening media, or may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
Example 1
Referring to fig. 1 to 4, a first embodiment of the present invention provides a method for identifying a fault of a power distribution room device based on infrared image analysis, including:
s1: and acquiring an infrared image of the power distribution room equipment, and performing image preprocessing on the infrared image.
Wherein, an infrared CCD detector is adopted to collect an infrared image; since there are many types of failures of electrical equipment and there is a certain difference between each type of failure, in order to locate the failure using infrared image data, it is necessary to segment an image, which is a picture divided into one or more regions using data such as grayscale, depth, texture, and color.
The image segmentation of the infrared data of the electrical equipment mainly has the following characteristics: firstly, infrared radiation signals are relatively weak, and if pictures are collected at a remote position, blurred pictures can be obtained; secondly, the infrared signal is disturbed by many external environmental factors, so that only the approximate shape of the object can be obtained generally; finally, there may be multiple electrical devices in the infrared image that may overlap, thereby affecting the accuracy of the analysis.
For these problems, the embodiment first performs noise reduction on the infrared image by using a mean filtering method; splicing the infrared images subjected to noise reduction processing through feature matching; and finally, segmenting the spliced infrared image by using an OSTU algorithm.
The image segmentation method comprises the following specific steps:
(1) the mean filtering is a linear filtering method for averaging pixel values of adjacent regions; the principle is that the target point is used as the center, and eight adjacent pixel values are calculated to obtain the pixel average value.
(2) In order to ensure the rotation invariance, the image is rotated, the feature point F (x, y) of the infrared image after the smoothing processing is defined, the main direction of the feature point F (x, y) is set as a coordinate axis, and the coordinates of the feature point after the rotation are as follows:
further, with the feature point as a center, calculating gradient values and directions of the region where the feature point is located, and dividing the gradient directions into 36 parts to establish a feature vector;
gradient value:
where L (x, y) is the size of the feature point.
The direction is as follows:
preferably, by establishing feature vectors at the feature points, the measurement of similarity in feature matching is facilitated; the feature vector not only contains information about feature points and neighboring pixels, but also can eliminate the influence of luminance variation on feature matching to some extent.
And further, performing feature matching on the smoothed images by using the feature vectors to complete image splicing.
(3) It should be noted that the OSTU algorithm uses the special property of image gray-scale values to segment and divide the image, and its principle is to measure the difference degree between the target threshold and the non-target threshold by calculating the variance between the two categories.
Defining X as an L-level gray image, and dividing all pixels in the image into a target class C0 and a non-target class C1 according to a threshold k, wherein the gray value range of all pixels in a C0 domain is [0, k-1], and the gray value range of all pixels in a C1 domain is [ k, L-1 ];
the proportion of the total area occupied by the pixels of C0 is:
the proportion of the total area occupied by the pixels of C1 is:
ω1＝1-ω0
wherein, PiThe area ratio of the ith pixel point to the C0 domain is represented as i, i is 0, 1, 2, …, m;
② define the maximum variance between classes C0 and C1 as:
δ2(k)＝ω0(μ-μ0)2+ω1(μ-μ1)2
wherein μ is an average gray level; mu.s0Is the variance of C0, and,
μ1is the variance of C1, μ1(k)＝1-μ0(k)。
S2: and identifying the power distribution room equipment in the infrared image based on the convolutional neural network.
It should be noted that the common types of electrical distribution room equipment mainly include insulators, windings, bushings, radiators, conservators, lead connectors, and switches.
The present embodiment classifies these types using a convolutional neural network whose basic structure includes an input layer, a convolutional layer, an activation layer, a pooling layer, a full-link layer, and an output layer.
The identification steps are as follows:
(1) the picture processed in S1 is input to a convolutional neural network through an input layer, and the size of the picture is normalized to 32 × 32.
(2) Convolutional layer
Is convolved as follows:
wherein M isjI is the size of the input image, i is the step size, j is the height of the image, l is the width of the image,
is the number of the convolution kernels and is,
is the size of the convolution kernel and,
is the number of channels.
(3) The activation layer is a non-linear mapping of the convolution result, specifically, activation is performed using ReLU, and the activation function is f (x) ═ max (0, h), when h <0 is input, the output is 0, and when h >0, the output is h; the activation function causes the network to converge more quickly.
(4) The pooling was performed by maximum pooling, as shown in FIG. 2.
(5) And simulating a full-connection layer by using a convolution mode, and performing convolution twice in total.
(6) And outputting the image characteristics of the power distribution room equipment through the output layer.
(7) And identifying by comparing the image characteristic information.
S3: and carrying out image registration on the infrared image through the combination device.
Assuming that A is an image with correct equipment position and B is an image to be registered;
taking any point Q (m, n) on the B diagram and any point R (u, v) on the A diagram, and respectively substituting the points into a registration equation:
then solving the equation to obtain a and b;
further, after obtaining the equation, traversing all the pixel points (m) of the B pictured，nd) Respectively substituting the pixel points into an equation to obtain the position coordinates (u) of all the registered pixel pointsd，vd)。
S4: and constructing a depth confidence network model, and carrying out fault diagnosis on the registration image.
It should be noted that the Deep Belief Network (DBN) is composed of multiple layers of Restricted Boltzmann Machines (RBMSs), the RBMs are probabilistic graphical models of stochastic neural networks, and the output of neurons thereof has only two states: active and inactive, and each output state has a probability of being determined; the network structure of RBMs is shown in fig. 3, which includes a visible layer and a hidden layer.
The last layer of the DBN is logistic regression, and the network structure is shown in fig. 4; the training process of the network comprises two processes: pre-training and fine-tuning.
Specifically, (1) pre-training: pre-training RBMs layer by layer under the unsupervised condition, and deeply mining hidden characteristic information of data, wherein each training epoch only trains one layer of RBMs; (2) fine adjustment: training and stacking the next layer of RBMs, combining the label with the sample for use, and then performing supervised adjustment by adopting back propagation to realize fault classification;
updating network parameters of the deep confidence network model by adopting a back propagation strategy, and defining a cost function as follows:
wherein E is the average square error, N is the number of hidden elements,
and XiRespectively representing the output of the output layer and the ideal output, i being the sample index, (W)l,bl) Representing the weights to be learned and the bias parameters at level l.
Further, the weight and the bias parameters are updated by adopting a gradient descent method:
where λ is learning efficiency.
Part of the code for the DBN model training is as follows:
function DBN＝DBNsetup(DBN,x,opts)
n ═ size (x, 2); # n is the characteristic dimension of a single sample.
-dbn. sizes ═ n, dbn. sizes ]; sizes are dimensions of RBMs.
for 1: numel (dbn. sizes) -1# numel (dbn. sizes) returns the number of elements in dbn. sizes.
Rbms { u }. alpha ═ ops.alpha; # initializes the learning rate of the RBMs.
DBN.RBMs{u}.momentum＝opts.momentum；
Rbms { u }. W ═ zeros (dbn. sizes (u +1), dbn. sizes (u)); # all initial values are 0.
DBN.RBMs{u}.vW＝zeros(DBN.sizes(u+1),DBN.sizes(u))；
Rbms { u }. b ═ zeros (dbn. sizes (u), 1); bias values of # display layer were all 0 at the initial value.
Rbms { u }. vb ═ zeros (dbn. sizes (u), 1); # the first RBMs is 100 and the second RBMs is 100.
DBN.RBMs{u}.c＝zeros(DBN.sizes(u+1),1)；
DBN.RBMs{u}.vc＝zeros(DBN.sizes(u+1),1)；
Still further, the registered image is input to the model, and the type of failure of the device is output by the model.
Preferably, errors and uncertainties due to manual reading of infrared images can be avoided by using a deep belief network to identify malfunctioning devices.
Example 2
In order to verify and explain the technical effects adopted in the method, the embodiment respectively selects the methods of threshold segmentation, region extraction, edge detection, forward neural network and support vector machine and adopts the method to carry out comparison test, and compares the test results by means of scientific demonstration to verify the real effect of the method.
The threshold segmentation has low image segmentation quality, the accuracy of region extraction and edge detection methods on target extraction is low, and the requirement on image quality is high.
In order to verify that the method has higher identification precision and higher segmentation quality relative to threshold segmentation, region extraction and edge detection methods are adopted in the embodiment to respectively identify and compare the acquired infrared images of the device.
(1) Each power device is represented as a class, and the faulty component is respectively labeled as another class, and the Threshold Segmentation (TS), Area Extraction (AE) and Edge Detection (ED) and the method respectively perform computational comparison on Pixel Accuracy (PA), Mean Pixel Accuracy (MPA), Mean cross over unit (MIoU) and Weighted cross over unit (iofwu) to measure the processing quality of each method on the image, and the computational formulas are respectively as follows:
wherein K is a prediction number; p is a radical ofijIs the number of pixels belonging to class i, but predicted to be class j; p is a radical ofjiIs the number of pixels belonging to class j, but is predicted to be class i; p is a radical ofiiIs the number of pixels belonging to the category i and is predicted as the number of pixels of the category i.
The calculation results are shown in table 1.
Table 1: threshold segmentation, region extraction and edge detection methods and a comparison table of results of the method for calculating accuracy are used.
It can be seen from the above table that the method improves the segmentation quality of PA and MPA by more than 5% compared with the conventional method.
(2) Performance assessment of fault status classification of equipment is mainly performed using recall and accuracy:
the recall ratio is as follows: recall TP/(TP + FN)
The precision ratio is as follows: precision TP/(TP + FP)
Wherein, tp (true positive): judging the correct number in the samples judged as positive; FP (false Positive): judging the number of errors in the samples judged as positive; tn (true negative): judging the correct number in the samples judged as negative; fn (false negative): the number of errors is determined for the samples determined to be negative.
The extracted hidden information is used as input to the DBN network and 7 types of devices are used as output.
Table 2 shows the classification results of DBN, SVM (support vector machine) and BPNN (Back prediction Neural Network).
Table 2: and different types of classification performance comparison tables of the power distribution room equipment.
As can be seen from the table, the method has high accuracy in transformer fault classification, and performs best on recall rate and accuracy, and the classification effect of the BPNN is the worst.
In order to better verify the effectiveness of the method for identifying the equipment fault, the whole process of analyzing the equipment fault of the power distribution room based on the infrared image is tested. First, image recognition is performed on the input RGB image, then the infrared image is used for image registration, and finally, the trained deep confidence network is used to determine the fault type in conjunction with the device temperature change.
Fig. 6 and 7 are graphs of actual results of identifying equipment faults by using the method, and as can be seen from the graphs, the method can accurately extract the targets of the power distribution room equipment existing in the graphs, and by combining infrared images, image registration can be more accurately realized.
In the apparatus extracted in fig. 6, the lowest temperature detected was 39.02 ℃ and the highest temperature was 39.28 ℃; the difference between the two is less than 1%, and the current normal state of the equipment can be judged according to the analysis result of the deep belief network; in the device proposed in fig. 7, the lowest temperature is found to be 39.42 ℃, the highest temperature is found to be 41.68 ℃, the difference between the two temperatures exceeds 5%, and meanwhile, the result output by the deep confidence network model also determines that the device is in an abnormal operation mode, and at this time, an early warning signal needs to be sent out in time to inform a worker to carry out further device operation state investigation.
It should be noted that the above-mentioned embodiments are only for illustrating the technical solutions of the present invention and not for limiting, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, which should be covered by the claims of the present invention.
Claims (10)
1. A power distribution room equipment fault identification method based on infrared image analysis is characterized in that: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
acquiring an infrared image of power distribution room equipment, and performing image preprocessing on the infrared image;
identifying power distribution room equipment in the infrared image based on a convolutional neural network;
performing image registration on the infrared image by combining the equipment;
and constructing a depth confidence network model, and carrying out fault diagnosis on the registration image.
2. The distribution room equipment fault identification method based on infrared image analysis according to claim 1, characterized in that: the image pre-processing includes the steps of,
carrying out noise reduction processing on the infrared image by using a mean value filtering method;
splicing the infrared images subjected to the noise reduction treatment through characteristic matching;
and dividing the spliced infrared image by using an OSTU algorithm.
3. The distribution room equipment fault identification method based on infrared image analysis according to claim 2, characterized in that: the splicing comprises the steps of splicing the two materials,
defining a characteristic point F (x, y) of the infrared image after the noise reduction processing, enabling the main direction of the characteristic point F to be a coordinate axis, and rotating the coordinate of the characteristic point:
taking the feature point as a center, and obtaining a feature vector by calculating the gradient value and the direction of the region where the feature point is located;
and performing the feature matching by using the feature vector.
4. The distribution room equipment fault identification method based on infrared image analysis according to claim 3, characterized in that: the gradient values may include, for example,
wherein L (x, y) is the ratio of the feature points.
5. The distribution room equipment fault identification method based on infrared image analysis according to claim 3 or 4, characterized in that: the directions include that the direction of the light beam comprises,
6. the distribution room equipment fault identification method based on infrared image analysis according to any one of claims 2, 3 and 4, characterized by comprising the following steps: the OSTU algorithm includes the steps of,
defining X as an L-level gray image, and dividing all pixels in the image into a target class C0 and a non-target class C1 according to a threshold k, wherein the gray value range of all pixels in a C0 domain is [0, k-1], and the gray value range of all pixels in a C1 domain is [ k, L-1 ];
the proportion of the total area occupied by the pixels of C0 and C1 is respectively as follows:
ω1＝1-ω0
wherein, PiThe area proportion of the ith pixel point in the C0 domain is that i is 0, 1, 2, …, m;
defining the maximum variance between the C0 and C1 classes as:
δ2(k)＝ω0(μ-μ0)2+ω1(μ-μ1)2
wherein μ is the average gray scale, μ0Is the variance of C0, μ1Is the variance of C1.
7. The distribution room equipment fault identification method based on infrared image analysis according to any one of claims 1, 2 and 4, characterized by comprising the following steps: the convolutional neural network comprises a convolutional neural network comprising,
convolutional layer
The calculation formula of (a) is as follows:
wherein M isjFor the size of the input image, i is the step size, j is the height of the image, l is the width of the image,
is the number of the convolution kernels and is,
is the size of the convolution kernel and,
is the number of channels.
8. The distribution room equipment fault identification method based on infrared image analysis according to claim 7, characterized in that: the image registration includes the registration of the images,
assuming that A is an image with the correct position of the equipment and B is an image to be registered;
taking any point Q (m, n) on the B diagram and any point R (u, v) on the A diagram, and respectively substituting the points into a registration equation:
and obtaining the values of a and B by solving the registration equation, traversing all pixel points of the B image and substituting the pixel points into the registration equation, and further obtaining all registration pixel points.
9. The distribution room equipment fault identification method based on infrared image analysis according to any one of claims 3, 4 and 8, characterized by comprising the following steps: the deep belief network model includes, for each of the plurality of network nodes,
pre-training: pre-training the Boltzmann machine layer by layer under the unsupervised condition, and deeply mining hidden characteristic information of data;
fine adjustment: training and stacking the next layer of the Boltzmann machine, combining the labels with the samples for use, and performing supervised adjustment by adopting back propagation to realize fault classification.
10. The distribution room equipment fault identification method based on infrared image analysis according to claim 9, characterized in that: the counter-propagation includes the fact that,
updating the network parameters of the deep belief network model by adopting the back propagation strategy, and defining a cost function as follows:
wherein E is the average square error, N is the number of hidden elements,
and XiRespectively representing the output of the output layer and the ideal output, i being the sample index, (W)l,bl) Representing the weights to be learned and the bias parameters at level l.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011250191.9A CN112419243B (en) | 2020-11-11 | 2020-11-11 | Power distribution room equipment fault identification method based on infrared image analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011250191.9A CN112419243B (en) | 2020-11-11 | 2020-11-11 | Power distribution room equipment fault identification method based on infrared image analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112419243A true CN112419243A (en) | 2021-02-26 |
| CN112419243B CN112419243B (en) | 2023-08-11 |
Family
ID=74781401
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011250191.9A Active CN112419243B (en) | 2020-11-11 | 2020-11-11 | Power distribution room equipment fault identification method based on infrared image analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112419243B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116050807A (en) * | 2023-03-30 | 2023-05-02 | 东莞正大康地饲料有限公司 | Feed production equipment defect identification method and system based on video monitoring inspection |
| CN116797854A (en) * | 2023-08-18 | 2023-09-22 | 常州思瑞电力科技有限公司 | Abnormality monitoring learning method based on photovoltaic box transformer substation monitoring system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109669087A (en) * | 2019-01-31 | 2019-04-23 | 国网河南省电力公司 | A kind of method for diagnosing fault of power transformer based on Multi-source Information Fusion |
| US20190243735A1 (en) * | 2018-02-05 | 2019-08-08 | Wuhan University | Deep belief network feature extraction-based analogue circuit fault diagnosis method |
| CN111144325A (en) * | 2019-12-28 | 2020-05-12 | 广东电网有限责任公司 | Fault identification and positioning method, device and equipment for power equipment of transformer substation |
-
2020
- 2020-11-11 CN CN202011250191.9A patent/CN112419243B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190243735A1 (en) * | 2018-02-05 | 2019-08-08 | Wuhan University | Deep belief network feature extraction-based analogue circuit fault diagnosis method |
| CN109669087A (en) * | 2019-01-31 | 2019-04-23 | 国网河南省电力公司 | A kind of method for diagnosing fault of power transformer based on Multi-source Information Fusion |
| CN111144325A (en) * | 2019-12-28 | 2020-05-12 | 广东电网有限责任公司 | Fault identification and positioning method, device and equipment for power equipment of transformer substation |
Non-Patent Citations (3)
| Title |
|---|
| FENGBINGCHUN: "图像配准算法", 《CSDN》 * |
| 杨博雄 等: "深度学习理论与实践" * |
| 甘伟焜: "基于红外图像的变压器图像处理方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116050807A (en) * | 2023-03-30 | 2023-05-02 | 东莞正大康地饲料有限公司 | Feed production equipment defect identification method and system based on video monitoring inspection |
| CN116050807B (en) * | 2023-03-30 | 2023-06-23 | 东莞正大康地饲料有限公司 | Feed production equipment defect identification method and system based on video monitoring inspection |
| CN116797854A (en) * | 2023-08-18 | 2023-09-22 | 常州思瑞电力科技有限公司 | Abnormality monitoring learning method based on photovoltaic box transformer substation monitoring system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112419243B (en) | 2023-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109118479B (en) | Capsule network-based insulator defect identification and positioning device and method | |
| CN112199993B (en) | Method for identifying transformer substation insulator infrared image detection model in any direction based on artificial intelligence | |
| CN106846316B (en) | Automatic identification method for typical defect images in GIS | |
| CN111784633B (en) | Insulator defect automatic detection algorithm for electric power inspection video | |
| CN110147777B (en) | Insulator category detection method based on deep migration learning | |
| CN111611874B (en) | Face mask wearing detection method based on ResNet and Canny | |
| CN113205063A (en) | Visual identification and positioning method for defects of power transmission conductor | |
| CN113344475B (en) | Transformer bushing defect identification method and system based on sequence modal decomposition | |
| CN112419243A (en) | Power distribution room equipment fault identification method based on infrared image analysis | |
| CN113554526A (en) | Fault early warning method and device for power equipment, storage medium and processor | |
| CN116612098B (en) | Insulator RTV spraying quality evaluation method and device based on image processing | |
| WO2024021461A1 (en) | Defect detection method and apparatus, device, and storage medium | |
| CN115861210B (en) | Transformer substation equipment abnormality detection method and system based on twin network | |
| CN111950566A (en) | Rotation-invariant HOG infrared image power equipment identification method | |
| CN111428772B (en) | Photovoltaic system depth anomaly detection method based on k-nearest neighbor adaptive voting | |
| Liu et al. | Ground-based cloud classification using weighted local binary patterns | |
| Lu et al. | Thermal Fault Diagnosis of Electrical Equipment in Substations Based on Image Fusion. | |
| CN117670820B (en) | Plastic film production defect detection method and system | |
| CN114445369A (en) | Contact net sectional insulator identification method and device based on 3D imaging technology | |
| CN116485802B (en) | Insulator flashover defect detection method, device, equipment and storage medium | |
| CN116188445A (en) | Product surface defect detection and positioning method and device and terminal equipment | |
| CN115937492A (en) | Transformer equipment infrared image identification method based on feature identification | |
| CN115631154A (en) | Power equipment state monitoring and analyzing method and system | |
| CN114220084A (en) | Distribution equipment defect identification method based on infrared image | |
| CN114373205A (en) | Face detection and recognition method based on convolution width network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |